life-history variation in a tropical cooperative bird: ecological and … · 2017. 3. 16. ·...
TRANSCRIPT
Life-history variation in a tropical cooperative bird: Ecological
and social effects on productivity
Nataly Alexandra Hidalgo Aranzamendi
M.Sc. and B.Sc.
A thesis submitted for the degree of Doctor of Philosophy at
Monash University in 2016
School of Biological Sciences
ii
New page
Copyright notice
© N. A. Hidalgo Aranzamendi (2016). Except as provided in the Copyright Act 1968, this thesis
may not be reproduced in any form without the written permission of the author.
iii
Abstract
Reproduction and survival are two fundamental traits that define an individuals’ lifetime. Life-
history variation within individuals can be related to individual differences on balancing
reproduction and survival against the constraints imposed by the physical and environmental
conditions they experience. Since individuals vary widely within and between populations in
respect to life-history traits, this variation transcends individuals, and life-history strategies differ
between species. Understanding the selective pressures that affect individuals will offer an insight
on what limits each species’ life-history strategy. Here I focus on the ecological and social forces
that affect reproduction of the cooperative breeding purple-crowned fairy-wren Malurus coronatus,
a tropical species that is located at the slow-life history end of the continuum. Using ten years of
data from individual longitudinal stories, I analyse many aspects of reproduction: what are the
environmental cues that trigger timing of breeding, how much do these environmental variables
alone or in combination with social and other ecological conditions influence the final outcome of
reproduction and what strategies do pairs use to minimise losses produced by external factors. I
demonstrate that Malurus coronatus has a fine-tuned mechanism to time breeding cycles with
rainfall and that rainfall also determines the quantitative output of successful nest attempts.
However, means and extremes of rainfall and temperature have different antagonistic effects that
directly or indirectly decrease reproductive success, affecting reproductive success by flooding,
increased predation, reduced egg hatchability or lower offspring survival. To overcome such losses,
pairs should invest in nest defence and brood rearing, and I test whether this could be improved as a
result of long-term partnerships. However, I found that retaining the same partner does not increase
productivity. Motivated by this lack of benefits of staying together, I analysed if an adaptive
response was to change partners (divorce) if opportunities are available. Divorce is driven by
females taking breeding positions in higher quality territories, however this does not lead to
iv
immediate reproductive improvements. More generally, high quality territories (with more
vegetation cover) are consistently identified as key factors associated with increased productivity by
decreasing the likelihood of nest failure and recruiting more fledglings, which confirms previous
research on this species and highlights the importance of “real estate”. All findings in Malurus
coronatus could possibly be commonplace in other tropical species that occupy year-round
territories and are exposed to similar environmental selective pressures. These findings increase
what is known for tropical species with a slow-life history strategy, data that is currently scarce.
Moreover, my research highlights some potential threats of climate change, in which adaptive
responses might not be enough to ameliorate negative impacts due to the slow-pace of reproduction
in this species. The importance of habitat quality on reproductive success suggest that the
preservation of healthy habitat would be crucial to guarantee this species survival.
v
Declaration
This thesis contains no material which has been accepted for the award of any other degree or
diploma at any university or equivalent institution and that, to the best of my knowledge and belief,
this thesis contains no material previously published or written by another person, except where due
reference is made in the text of the thesis.
vi
)
Publications during enrolment
Thesis including published works General Declaration
I hereby declare that this thesis contains no material which has been accepted for the award of any
other degree or diploma at any university or equivalent institution and that, to the best of my
knowledge and belief, this thesis contains no material previously published or written by another
person, except where due reference is made in the text of the thesis.
This thesis includes one original paper published in a peer reviewed journal and three
unpublished publications. The core theme of the thesis is the Ecological and Social factors that
affect productivity in a tropical bird. The ideas, development and writing up of all the papers in the
thesis were the principal responsibility of myself, the candidate, working within the School of
Biological Sciences of Monash University under the supervision of Dr Anne Peters.
The inclusion of co-authors reflects the fact that the work came from active collaboration
between researchers and acknowledges input into team-based research.
In the case of all chapters my contribution to the work involved the following:
Thesis
chapter
Publication title Publication status Nature and extent (%) of
students contribution
1 Fast response to rainfall connects
the timing of breeding to food
abundance and reproductive output
in a tropical bird
In preparation The candidate collected and
interpreted most of the data
(90 %). Primary author
vii
2 Contrasting effects of climate
means and extremes on nest failure
and productivity in an endangered
tropical bird
In review The candidate collected and
interpreted most of the data
(90 %). Primary author
3 Limited benefits of mate retention
for productivity in a tropical
passerine
In preparation The candidate collected and
interpreted most of the data
(90 %). Primary author
4 Incest avoidance, extra-pair
paternity and territory quality drive
divorce in a year-round territorial
bird
Published The candidate collected and
interpreted most of the data
(90 %). Primary author
I have renumbered sections of submitted or published papers in order to generate a consistent
presentation within the thesis.
Student signature: Date: 15-10-2016
The undersigned hereby certify that the above declaration correctly reflects the nature and extent of
the student and co-authors’ contributions to this work.
Main Supervisor signature: Date: 15-10-2016
viii
Acknowledgements
I would like to thank my supervisory panel Anne Peters, Michelle L. Hall and Paul Sunnucks. I was
privileged to work with Anne Peters, she always provided guidance, support, feedback and all our
discussions provided interesting results. She always shared my enthusiasm when doing fieldwork
and provided vital advices while discussing interesting results or brainstorming possible
hypotheses. Her knowledge, example and support help me become a better scientist. I also thank
Michelle for her constant and accurate feedback, her deep knowledge of fairy-wrens and in
particular our system, was key to advance during all the stages of my PhD. Following her steps and
filling her shoes conducting fieldwork was a big personal challenge that I took with enthusiasm.
I was privileged to be surrounded by a remarkable group of co-authors: Kaspar Delhey,
Sjouke A. Kingma, Paul Sunnucks and Martijn van de Pol. Your constant and willingly available
help, constructive criticism and positive feedback help me enhance the quality of this work.
This work would have not been possible without the financial support of numerous
institutions. I thank the School of Biological Sciences at Monash University, the Australian
Research Council and the Australian Wildlife Conservancy. Additional support was provided by the
Holsworth Wildlife Endowment and the Stuart Leslie Bird Research Award. Moreover my PhD was
supported by a Monash Research and Graduate Scholarship (MRGS) and an International
Postgraduate Research Scholarship (IPRS).
The success of my fieldwork would have not been possible without the people that
voluntarily help me capturing and observing the birds. All these tasks were done with enthusiasm
ix
despite the hard field conditions and I was lucky to count with such a remarkable crew of field
companions. My gratitude to Katie King, Rowan Jacques-Hamilton, Anne Aulsebrook, Lisa
Mandeltort, Michael Hansen, Stephanie McClelland, Chris Smith, Marie Fan, Niki Teunissen and
Mattia Pierangelini. My safety and the success of every field visit would not have been possible
without the support of all the staff of Australian Wildlife Conservancy in Mornington Sanctuary. I
am particularly grateful to Sarah Legge, Katherine Tuft and Hugh McGregor for being always
helpful with my work and enthusiastic about all aspects of my research. The times that I spend there
were unforgettable and enjoyable and naming everyone that help me and became a friend during my
stay at Mornington would be endless, but my gratitude goes to them all.
My PhD experience was complete with the constant support and companionship of my
many PhD friends. I thank all the members of my lab, my housemates and all friends with whom I
shared great moments interchanging questions and going through all good and not-so-good times
during lunch, while sharing a beer or by just emailing encouraging words. Name you all would be
never-ending but you should know that our constant science and life dialogue was vital to succeed
and finalize my work.
Most importantly, I would have not adventured in such journey without the complete
support of my family: Soledad, Edgardo, Cynthia and Mattia. Your constant support and love was
my driver to achieve all goals. I thank you for understanding the times that I had to spend away
doing what I like the most, looking for birds, and for always believing in me.
x
Table of contents
Preface………………………………………………………………………………………….…1
Introduction ………………………………………………………………......……………….….2
Chapter 1. Fast response to rainfall connects the timing of breeding to food abundance and
reproductive output in a tropical bird……………………………………………………………17
Chapter 2. Contrasting effects of climate means and extremes on nest failure and productivity in
an endangered tropical bird…………………………………………………………………...…51
Chapter 3. Limited benefits of mate retention for productivity in a tropical passerine………...82
Chapter 4. Incest avoidance, extra-pair paternity and territory quality drive divorce in a year-
round territorial bird……………………………………………………………………………111
General discussion……………………………………………………………………………...153
1
Preface
This thesis contains an introduction, four data chapters and a final discussion. Each data
chapter constitutes an independent manuscript and its content is formatted according to the
requirements of a particular journal. Amendments have been done in format of references,
tables and figures in order to combine all chapters. Since each chapter is intended as an
independent publication, some repetition of information is found. Chapter four has been
published in the journal Behavioral Ecology (Hidalgo Aranzamendi et al. 2016) and it is
referred to as Chapter 4 throughout the document. Chapter two has been submitted to the
journal Functional Ecology (Hidalgo Aranzamendi et al. under review) and it is cited as
Chapter 2 throughout. The other 2 chapters are in preparation for submission and referred to
by their respective number throughout the document.
This thesis work is my own. All contributors are listed as co-authors including supervisors.
2
Introduction
Background
Reproduction and survival are fundamental life-history traits that outline an individual’s
lifetime (Williams 1966, Stearns 1992, Roff 1993). Selection is constrained because
reproduction and survival cannot be maximized simultaneously; instead both aspects require
a constant trade-off in allocation of resources and eventually each individual does this
differently throughout its life (Williams 1966, Partridge and Harvey 1985, Hille and Cooper
2015). Since some individuals are better than others in finding territories, partners or just
surviving in general, such differential allocation is reflected in a large variation between
individuals’ reproductive output (van de Pol 2006). Individuals vary widely within and
between populations with respect to life-history traits, and ultimately this variation transcends
individuals, and life-history strategies differ between species (Stearns 1992, Ricklefs and
Wikelski 2002).
The pace-of-life of a species is a composite of several traits that are the product of life-history
trade-offs that individuals face (Hille and Cooper 2015). It has been suggested that each
species falls along a continuum going from a slow pace-of-life to a fast pace-of-life strategy
(Wikelski et al. 2003, Reale et al. 2010). In birds, most of our understanding of life-history
processes comes from the northern hemisphere, where most species fall at the end of the fast
continuum and have a reproduce-fast, die-young strategy (Martin 1996). In contrast, species
with a slow pace-of-life have: small clutch sizes (Martin 1995b, Stutchbury and Morton
2001), longer incubation and nestling periods (Ricklefs 1969, 1976, Styrsky et al. 2005),
prolonged parental care (Russell 2000, Hille and Cooper 2015), low growth rates (Wiersma et
al. 2007, Reale et al. 2010), high philopatry (Reale et al. 2010), high juvenile and adult
survival (Ghalambor and Martin 2001), late sexual maturation (Russell 2000) and delayed
3
senescence (Møller, et al. 2007). It is believed that most tropical species are found closer to
the slow pace-of-life end (Martin 1996, Stutchbury and Morton 2008, Møller, and Liang
2013). Since reproduction at the individual level eventually determines population dynamics,
and thereby affects a species’ evolutionary potential and its conservation, understanding the
slower pace-of-life of tropical species constitutes a major contribution to a core aspect of life-
history theory (Roff 1993). To understand this at the species level, and its implications in
population biology, it is important to analyse selective pressures that constrain trade-offs
between traits influencing productivity at the individual level.
The slow pace-of-life strategy of tropical species has been attributed to several selective
pressures that include: high rates of nest predation (Martin and Briskie 2009), lack of a
pronounced peak in food abundance, reduced seasonality and large unpredictability of the
environment (Ricklefs and Wikelski 2002), high density of predators (Martin 1995b, 1996),
diseases (Reale et al. 2010) (ecological drivers) and relationships within social and family
groups (Ricklefs and Wikelski 2002) (social drivers). Although these factors potentially
explain life-history traits of a slow pace-of-life, no consensus has emerged concerning the
mechanisms that constrain time and energy allocation to life-histories (Ricklefs and Wikelski
2002) and this is mostly due to lack of breeding biology data on tropical species (Martin
1996, 2015, Russell 2000, Stuchtbury et al. 2007, Møller, and Liang 2013).
Tropical cooperative breeders offer the opportunity to test several hypotheses proposed to
explain productivity of slow-paced species. Cooperatively breeding birds, a group
disproportionately represented in the southern hemisphere and in the tropics, are long-lived
animals in which trade-offs between reproduction and survival are played out at a particularly
slow pace (Cockburn 1998, Koenig and Dickinson 2004, Feeney et al. 2013). Ecological
factors, especially environmental variation, has been proposed to be an important driver
explaining the occurrence of cooperative breeding (Jetz and Rubenstein 2011). In here,
4
environmental uncertainty and resource availability (density of breeding sites and territory
quality) drive temporal variation and exert selection pressures on cooperation (Walters 1990,
Komdeur 1992, Jetz and Rubenstein 2011). Moreover, cooperative breeders delay
reproduction until they inherit a breeding position or a vacancy arises elsewhere (Koenig and
Dickinson 2004); this delay in reproduction is compensated by increased longevity
(Cockburn 1998, Hatchwell and Komdeur 2000). Since these species are found in highly
saturated environments surrounded by relatives, they face stringent constraints on movements
and pairing opportunities (Cockburn 1998, Hatchwell and Komdeur 2000). These
demographic characteristics combined, make them a good system to explore how ecological
or social effects determine variation on individuals’ decisions on reproduction that might
affect survival of those individuals (Stearns 1992, Roff 1993, Martin 1995a).
Thesis aim
In this thesis I study the purple-crowned fairy-wren, a cooperative breeder species found in
the monsoonal tropics of Australia, to disentangle how ecological and/or social factors
determine variation on individual productivity. This information will be relevant to identify
possible threats that jeopardize this bird, since this species has an endangered status.
For this, I used data from a long-term study (2005-2015), in which hundreds of individuals
were followed from birth until death. Over those ten years, social and ecological aspects of
purple-crowned fairy-wrens’ lives were recorded, such as, data for breeding performance,
dispersal, survival, territory quality, and environmental climatic factors.
5
Study species
The purple crowned fairy-wren Malurus coronatus belongs to the genus Malurus, an endemic
Australian-Papuan group (Rowley and Russell 1997). In this part of its distribution, purple-
crowned fairy-wrens exclusively occupy territories linearly aligned along water courses and
with dense vegetation of a palm-like species Pandanus aquaticus. (Rowley and Russell 1997,
Kingma 2009, 2011a). Here, each all-purpose territory is occupied year-round by one
dominant pair, which are the only ones that reproduce, and are sometimes accompanied by a
group of subordinates (helpers) varying from none to 10 (Kingma et al. 2009, 2010). Natal
dispersal in this species is female biased, although some females remain philopatric (Kingma
2011b). Movement away from watercourses and dispersal outside river vegetation, or
between catchments, has never been recorded for this species (Skroblin et al. 2014).
Unlike other congeners, purple-crowned fairy-wrens have low rates of extra-pair paternity
(<6%, Kingma et al. 2009, 2013). Breeding in this population has been recorded year-round,
with one marked peak over the wet season (Dec-Mar) and another variable peak at the end of
the dry season (Aug-Oct) (Peters et al. 2013).
The subspecies found at out study site, Malurus coronatus coronatus, has suffered declines in
their distribution and density as a result of degradation of riparian habitat (Skroblin 2012) and
to date, it is listed as endangered by the Environment Protection & Biodiversity Conservation
Act of Australia (2015).
Field work
The study population is located along Annie Creek and the Adcock River (17° 31’S 126°
6’E) in Mornington Wildlife Sanctuary, a protected area managed by the Australian Wildlife
Conservancy since 2004.
6
The long-term study of this population started in 2005 and until 2010, all birds were captured
and banded. During these years, breeding attempts were followed in detail (Kingma 2012).
All natal and breeder dispersal events were recorded by weekly censuses in the study site.
Since 2010 until 2015, two yearly censuses were conducted in May-June and November after
the main two breeding peaks.
All protocols were approved by the ethics committees of the School of Biological Sciences at
Monash University (BSCI/2011/28) and the Max Planck Institute for Ornithology under
national and international legislation for animal use in research. All research was approved by
the Western Australian Department of Environment and Conservation and by the Australian
Wildlife Conservancy (AWC).
Outline of thesis
Understanding which ecological and/or social factors determine variation on individual
productivity of purple-crowned fairy-wrens is the central theme of this thesis. Optimal
individual decisions include choosing the best time to breed, best age to breed, and best mate
to breed with, all of which influence reproductive outcomes (van de Pol 2006). Nevertheless,
individual’s strategies may be constrained, and in my thesis, I explore these aspects trying to
understand benefits of, and constraints on, individual strategies for reproduction, by testing
ecological and social factors that have been proposed as the main drivers of a slow pace-of-
life strategy. My thesis is partitioned into four data chapters, each of which discusses a
specific topic. The first part of the thesis (Chapter 1 and 2) is focussed on ecological drivers
of reproductive output, and the second part (Chapter 3 and 4) on social drivers that influence
reproductive output at the individual level. Finally, I discuss the main findings of all chapters
7
highlighting the main environmental factors that are affecting productivity in this species and
eventually how this influences population dynamics.
In Chapter 1, I explore the climatic factors that determine timing of breeding and the
consequences of timing on two aspects of productivity: clutch size variation and survival of
eggs and offspring. It has been proposed that the tropics lack strong seasonality when
contrasted with temperate regions, and cues that tropical species follow might differ from
recognizable cues in those striking seasonal environments (e.g. day length) (Hau et al. 1998,
2004, 2008). Therefore, identifying the cues for reproduction for tropical species would help
us to understand variation in components of the environment (e.g. food supply, predators)
that affect reproductive cycles (Grant et al. 2000, Wikelski et al. 2000, Schaper et al. 2012).
Such cues are likely to be tightly linked to the climatic characteristics of the region (Hau et al.
2008). Using a recently developed integrative modelling approach I analysed 685 nesting
records of 171 pairs in our study population and determined the relationship climate - food
availability - reproduction for the purple-crowned fairy-wren. In this chapter I showed how
climatic variation triggers reproduction. How climate affects the outcome of reproductive
attempts is topic of the next chapter.
Climate variation is generally assumed to be an important determinant of productivity and
survival in tropical species and cooperative breeders (Jetz and Rubenstein 2011, Oppel et al.
2013). In Chapter 2, I focus on how much the outcome of reproductive attempts (i.e. failure
or success in nest attempts) is determined by two aspects of climate variables (means and
extremes). Species responses to different components of the same climatic variable - means
versus extremes – are needed to understand effects of climate variability, and such responses
and the interaction with habitat quality have to be analyzed simultaneously. Such information
is critical, because it is likely that climate variables and thresholds affecting the biology of
tropical species cannot be extrapolated from findings in other latitudes. For instance, rainfall
8
constitutes a major climatic feature that defines seasonality in the tropics and a determinant of
the biological cycle of tropical species, rather than temperature, as is common in temperate
latitudes. Using an integrated approach, I combined almost 700 detailed nest records obtained
over 5 years with detailed information on climatic conditions they experienced and habitat
quality. I analyzed the effects of means and extremes of temperature and rainfall on the
outcome of all stages of the nesting cycle. This chapter highlights the complex effects of
environmental variation upon reproduction and how low reproductive success comes about in
tropical species.
Chapter 1 and 2 focussed on the major ecological factors affecting reproduction (climate,
habitat quality, and predation) and their impact in this population. Since variation in
reproduction is constrained by trade-offs between survival and individuals’ investments in
reproduction (Roff 1993), it is necessary to consider factors at the individual level that will
also determine reproduction. In Chapter 3, I analysed how individual and social breeding
history affect productivity. Effects of the length of relationships and breeding experience on
reproduction have been largely explored in different life-histories (Black 2001, Naves et al.
2007). Species establishing long-term pair bonds are predicted to do so, to increase
reproductive success over time (Black 2001) and that improvement occurs not only with
breeding experience per se, but also with increasing pair bond duration (Martin 1995a,
Fowler 1995). The effects of pair bond duration on productivity have been little explored in
species from tropical latitudes, despite the fact that several tropical species form such long-
term pair bonds (Stutchbury and Morton 2001, Tobias et al. 2016). I explore the effects of
pair bond duration on reproduction, using the breeding history of 194 pairs with known pair
bond duration and known breeding experience. The benefits of long pair bonds might be
absent or constrained in tropical cooperative species and explained by other ecological
9
factors, because cooperative breeders also experiment saturated environments with large
constraints in social partner choice.
The adaptive benefits or limitations that social partner choice have on successful reproduction
can be also analysed by studying the paths of formation and dissolution of relationships. In
Chapter 4, I focus in the causes and consequences of divorce. Despite the obvious benefits
of staying in a long-lasting pair bond, divorce is common in monogamous species
(Choudhury 1995). Only recently, it has become accepted that this behaviour is an adaptive
strategy to improve fitness (either reproduction or survival) (Culina et al. 2014). For species
that maintain pair bonds year-round, the limited availability of breeding positions presumably
constrains the opportunities for initial mate choice as well as partner change, but the
ecological and demographic drivers of divorce remain largely unknown. Investigating
divorce in purple-crowned fairy-wrens, allowed me to test several hypotheses in relation to
the full range of ecological and social external covariates and the impact that divorce has on
reproductive success of individuals.
References
Black, J. M. 2001. Fitness consequences of long-term pair bonds in barnacle geese:
Monogamy in the extreme. Behavioral Ecology 12: 640-645.
Cockburn, A. 1998. Evolution of Helping Behavior in Cooperatively Breeding Birds. Annual
Review of Ecology and Systematics 29: 141-177.
Choudhury, S. 1995. Divorce in birds: A review of the hypotheses. Animal Behaviour 50:
413-429
Culina, A., R. Radersma and B. C. Sheldon. 2014. Trading up: the fitness consequences of
divorce in monogamous birds. Biological Reviews 90:1015-1034.
10
Environment Protection & Biodiversity Conservation Act of Australia. 2015. Threatened
species scientific committee. Malurus coronatus coronatus 31/10/2015.
Feeney, W. E., I. Medina, M. Somveille, R. Heinsohn, M. L. Hall, R. A. Mulder, J. A. Stein,
R. M. Kilner and N. E. Langmore. 2013. Brood Parasitism and the Evolution of
Cooperative Breeding in Birds. Science 342: 1506-1508.
Fowler, G. S. 1995. Stages of Age-Related Reproductive Success in Birds: Simultaneous
Effects of Age, Pair-Bond Duration and Reproductive Experience. American Zoologist
35: 318-328.
Ghalambor, C. and T. E. Martin. 2001. Fecundity-survival trade-offs and parental risk-taking
in birds. Science 292: 494-499
Grant, P. R., B. R. Grant, L. F. Keller and K. Petren. 2000. Effects of El Niño events on
Darwin's Finch productivity. Ecology 81: 2442-2457.
Hall M. L. and A. Peters. 2009. Do male paternity guards ensure female fidelity in a duetting
fairy-wren? Behavioral Ecology 20: 222-228.
Hatchwell, B. J. and J. Komdeur. 2000. Ecological constraints, life history traits and the
evolution of cooperative breeding. Animal Behaviour 59: 1079-1086.
Hau, M., M. Wikelski, H. and J. C. Wingfield. 1998. A neotropical forest bird can measure
the slight changes in tropical photoperiod. Proceedings of the Royal Society of London B
265: 89-95.
Hau, M., M. Wikelski, H. Gwinner and E. Gwinner. 2004. Timing of reproduction in a
Darwin's finch: temporal opportunism under spatial constraints. Oikos 106: 489-500.
Hau, M., N. Perfito, I. T. Moore. 2008. Timing of breeding in Tropical Birds: Mechanisms
and Evolutionary Implications. Ornitologia Neotropical 19 (Suppl.): 39-59.
11
Hidalgo Aranzamendi, N., M. L. Hall, S. A. Kingma, P. Sunnucks and A. Peters. 2016. Incest
avoidance, extrapair paternity, and territory quality drive divorce in a year-round
territorial bird. Behavioral Ecology arw101.
Hille, S. M. and C. B. Cooper. 2015. Elevational trends in life histories: Revising the pace-of-
life framework. Biological Reviews 90:204–213.
Houston, W. A. 2013. Breeding cues in a wetland-dependent Australian passerine of the
seasonally wet-dry tropics. Austral Ecology 38: 617-626.
Jetz, W. and D. R. Rubenstein. 2011. Environmental Uncertainty and the Global
Biogeography of Cooperative Breeding in Birds. Current Biology 21: 72-78.
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology 9: 10.1186/1472-6785-9-15
Kingma S. A., M. L. Hall, E. Arriero and A. Peters. 2010. Multiple benefits of cooperative
breeding in Purple-crowned Fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79:757-768.
Kingma, S.A. 2011a. The evolution of social and genetic systems of Purple-crowned Fairy-
wrens. PhD Dissertation. Konstanz University, Germany.
Kingma, S. A., M. L. Hall and A. Peters. 2011b. Multiple Benefits drive helping behaviour in
a cooperatively breeding bird: An integrated analysis. The American Naturalist 177:
486:495.
Kingma, S. A., M. L. Hall and A. Peters. 2013. Breeding synchronization facilitates extrapair
mating for inbreeding avoidance. Behavioral Ecology 10.1093/beheco/art078
Koenig, W. and J. Dickinson. 2004. Evolution and Ecology of Cooperative Breeding in
Birds. Cambridge University Press
Komdeur, J. 1992. Importance of habitat saturation and territory quality for evolution of
cooperative breeding in the Seychelles warbler. Nature 358, 493–495.
12
Martin, K. 1995a. Patterns and mechanisms of age dependant reproduction and survival in
birds. American Zoologist 340-348.
Martin, T.E. 1995b. Avian life history evolution in relation to nest sites, nest predation and
food. Ecological Monographs 65:101-127
Martin, T. E. 1996. Life history evolution in tropical and south temperate birds: what do we
really know? Journal of Avian Biology 27: 263-273.
Martin, T. E. 2015. Age-related mortality explains life history strategies of tropical and
temperate songbirds. Science 349: 966-970.
Martin, T. E. and J. V. Briskie. 2009. Predation on dependent offspring: A review of the
consequences for mean expression and phenotypic plasticity in avian life history traits.
Annals of the New York Academy of Sciences 1168, 201–217.
Møller, A. P. 2007. Senescence in relation to latitude ad migration in birds. Journal of
Evolutionary Biology 20, 750–757.
Møller, A. P., W. Fiedler and P. Berthold. 2010. Effects of climate change in birds. Oxford
University Press.
Møller, A. P. and W. Liang. 2013. Tropical birds take small risks. Behavioral Ecology 24(1):
267-272.
Naves, L. C., E. Cam and J. Y. Monnat. 2007. Pair duration, breeding success and divorce in
a long-lived seabird: benefits of mate familiarity? Animal Behaviour 73: 433-444.
Oppel, S., G. M. Hilton, R. Allcorn, C. Fenton, A. J. Matthews and D. W. Gibbons. 2013.
The effects of rainfall on different components of seasonal fecundity in a tropical forest
passerine. Ibis 155: 464-475.
Partridge, L. and P. H. Harvey. 1985. The cost of reproduction. Nature 316:20.
Peters, A., S. A. Kingma and K. Delhey. 2013. Seasonal male plumage as a multi-component
sexual signal: insights and opportunities. Emu 113: 232-247.
13
Reale, D., D. Garat. M. M. Humphries, P. Bergeron, V. Careau and P. Montiglio. 2010.
Personality and the emergence of the pace-of-life syndrome concept at the population
level. Philosophical Transactions of the Royal Society B 10.1098/rstb.2010.0208
Ricklefs, R. E. 1969. An analysis of nesting mortality in birds. Smithsonian Contributions in
Zoology 9, 1–48.
Ricklefs, R. E. 1976. Growth rates of birds in the humid New World tropics. Ibis 118, 179–
207.
Ricklefs, R. E. and M. Wikelski. 2002. The physiology/life-history nexus. Trends in Ecology
and Evolution 17:462-469.
Roff, D. 1993. The evolution of life histories. Chapman & Hall. New York. USA.
Rowley, I. and E. Russell. 1997. Fairy-wrens and Grasswrens. Oxford University Press. New
York. USA.
Russell, E. M. 2000. Avian life histories: is extended parental care the southern secret? Emu
100: 377-399.
Schaper, S. V., A. Dawson, P. J. Sharp, P. Gienapp, S. P. Caro and M. E. Visser. 2012.
Increasing temperature, not mean temperature, is a cue for avian timing of reproduction. The
American Naturalist 179:e55-69.
Skroblin, A. 2012. Phylogeography and conservation biology of Purple-crowned Fairy-wren.
PhD Dissertation. Australian National University. Canberra, Australia.
Skroblin, A., A. Cockburn and S. Legge. 2014. The population genetics of the western
purple-crowned fairy-wren (Malurus coronatus coronatus), a declining riparian
passerine. Australian Journal of Zoology 10.1071/ZO13087
Stearns, S. C. 1992. The evolution of life histories. Oxford University Press. New York.
Stutchbury, B. J. M. and E. S. Morton. 2001. Behavioral Ecology of Tropical Birds.
Academic Press, San Diego.
14
Stutchbury, B. J. M. and E. S. Morton. 2008. Recent Advances in the Behavioral Ecology of
Tropical Birds. The Wilson Journal of Ornithology 120: 26-37.
Stutchbury, B. J. M., E. S. Morton and B. Woolfenden. 2007. Comparison of the mating
systems and breeding behavior of a resident and a migratory tropical flycatcher. Journal
of Field Ornithology 78: 40-49.
Styrsky, J. N., J. D. Brawn and S. K. Robinson. 2005. Juvenile mortality increases with
clutch size in a Neotropical bird. Ecology 86, 3238–3244.
Tobias, J. A., C. Sheard, N. Seddon, A. Meade, A. J. Cotton, S. Nakagawa and E. Datasets.
2016. Territoriality, social bonds, and the evolution of communal signaling in
birds. Frontiers in Ecology and Evolution 4: 74. doi: 10.3389/fevo.
van de Pol, M. 2006. State-dependent life-history strategies. A long-term study on
oystercatchers. PhD Thesis. University of Groningen.
van de Pol, M., D. Heg, L. W. Bruinzeel, B. Kuijper and S. Verhulst. 2006. Experimental
evidence for a causal effect of pair-bond duration on reproductive performance in
oystercatchers (Haematopus ostralegus). Behavioral Ecology 17: 982-991
van de Pol, M. and S. Verhulst. 2006. Age-Dependent Traits: A New Statistical Model to
Separate Within- and Between-Individual Effects. The American Naturalist 167: 766-
773.
Walters, J.R. 1990. Red-cockaded woodpeckers: A ‘primitive’ cooperative breeder. In
Cooperative Breeding in Birds: Long-Term Studies of Ecology and Behavior, P.B.
Stacey and W.D. Koenig, eds. (Cambridge: Cambridge University Press), 67–102.
Wiersma, P., A. Muñoz-Garcia, A. Walker and J. B. Williams. 2007. Tropical birds have a
slow pace-of-life. Proceedings of the National Academy of Sciences 104: 9340-9345.
Wikelski, M., M. Hau and J. C. Wingfield. 2000. Seasonality of reproduction in a neotropical
rain forest bird. Ecology 81: 2458-2472.
15
Wikelski, M., L. Spinney, W. Schelsky, A. Scheuerlein and E. Gwinner. 2003. Slow pace-of-
life in tropical sedentary birds: A common-garden experiment on four stonechat
populations from different latitudes. Proceedings of the Royal Society of London B:
Biological Sciences 270, 2383–2388.
Williams, G. C. 1966. Natural Selection, the Costs of Reproduction, and a Refinement of
Lack's Principle. The American Naturalist 100: 687-690.
16
PART B:
Declaration for Thesis Chapter 1
Declaration by candidate
In the case of Chapter 1, the nature and extent of my contribution to the work was the
following:
Nature of contribution Extent of contribution (%)
The key ideas, data collection, data analysis, lab work
and writing of the manuscript were my responsibility.
90
The following co-authors contributed to the work. If co-authors are students at Monash
University, the extent of their contribution in percentage terms must be stated:
The undersigned hereby certify that the above declaration correctly reflects the nature and
extent of the candidate’s and co-authors’ contributions to this work.
Candidate’s Signature Date: 15-10-2016
Main Supervisor’s Signature Date: 15-10-2016
Name Nature of contribution
Anne Peters Assistance with writing manuscripts, design of hypotheses, data
collection, interpretation of results, feedback on thesis
Michelle L. Hall Assistance with writing, data collection, interpretation of results.
Sjouke A. Kingma Assistance with writing manuscripts, data collection, interpretation
of results.
Martijn van de Pol Assistance with data analysis, interpretation of results.
17
Chapter 1
Fast response to rainfall connects the timing of breeding to food
abundance and reproductive output in a tropical bird
Nataly Hidalgo Aranzamendi1, Michelle L. Hall2,3, Sjouke A. Kingma3,4, Martijn van de
Pol5,6and Anne Peters1,3
1School of Biological Sciences, Monash University, Australia; 2School of BioSciences,
University of Melbourne, Australia; 3Max Planck Institute for Ornithology, Radolfzell,
Germany; 4Behavioural and Physiological Ecology, University of Groningen, The
Netherlands; 5Department of Evolution, Ecology & Genetics, Research School of Biology,
The Australian National University, Canberra, Australia;6Department of Animal Ecology,
Netherlands Institute of Ecology (NIOO-KNAW), Wageningen, the Netherlands
Abstract
Changes in climate are causing changes in the seasonal timing of many phenomena in the
natural world. Phenological mismatches due to global warming are well known from northern
temperate regions, but such effects are much less well understood in tropical and southern
regions. We used a new modelling approach to identify the critical climate signals that trigger
two biological responses: nest initiation by a tropical insectivorous bird, and increases in
arthropod abundance (representing its food). We then assessed how these critical climate
signals affected breeding success. Birds initiated nesting in response to cumulative rainfall in
the month prior to egg-laying, and no other climate variable predicted nest-initiation.
Arthropod abundance also increased in response to rainfall only, and a slower response meant
that the peak coincided with peak food demand of fledglings. The rapid nest-initiation response
of females to the same climate signal that triggers increases in arthropod abundance effectively
allows females to time their breeding to match optimal food supply for offspring. Indeed, the
18
amount of rainfall during the critical window predicted reproductive output. Knowledge about
the climate signals driving variation in food availability is essential to predict the effects of
climate change on breeding success globally.
Key words: phenology, tropical, ecological mismatch, multi-brooded, climate change
Introduction
The seasonal timing (phenology) of natural phenomena such as plant growth and animal
breeding are driven by seasonal changes in climate cues such as daylight, temperature and
precipitation (Hughes 2000, Schwartz 2003, Møller et al. 2008). Recent changes in global
climate are causing changes in the breeding phenology of taxa that reproduce seasonally. This
has been particularly noticeable for species living in northern temperate latitudes that use
increasing temperatures as a cue to start breeding (Crick et al. 1997, Hughes 2000, Visser et al.
2004, 2006b, 2008, Schaper et al. 2012). For instance, birds have advanced their laying dates
as a response of such increase (Both and Visser 2001, Dunn and Winkler 2010), and these
widely observed patterns constitute the best documented evidence of climate change on wild
animals (Root et al. 2003, Dunn and Winkler 2010).
The climate cues driving the seasonal timing of breeding in tropical birds are much less well
understood than those of the north-temperate zone, and little is known about the response of
tropical species to altered climate. This is mostly due to lack of detailed phenological data
and the fact that species-specific climatic factors that trigger breeding are poorly understood
(Charmantier and Gienapp 2014). Unlike species from temperate environments, the breeding
schedules of tropical birds can vary considerably. Breeding activity could be highly
predictable (e.g. Spotted Antbird, Hau et al. 1998), or breeding can be opportunistic with high
levels of flexibility (e.g. Zebra Finches or Galapagos Finches, Grant et al. 2000, Wikelski et
al. 2000, Hau et al. 2004). Although day length changes seasonally in near-equatorial
19
habitats, the amplitude of this variation is much reduced closer to the equator (Hau et al.
2008). Similarly, ambient temperature in near-equatorial tropical habitats is subjected to less
variation (Wikelski et al. 2000, Hau et al. 2004), thus these factors are unlikely to be the main
cues for breeding (but see Hau et al. 1998). Instead, rainfall is likely to be important (Hau et
al. 2008).
Rainfall, especially where it defines seasonality (monsoonal tropics) or is the limiting
resource (arid zone), has been proposed as the most likely strong determinant of breeding
calendars for many tropical species (Wingfield et al. 1992, Hau 2001, Shine and Brown
2008), predicting food peaks better than day length (Grant et al. 2000). Although tropical
rainfall might have a mostly predictable annual schedule (Hau et al. 2008), it has wide local
variability and relatively low predictability of frequency and intensity (Shine and Brown
2008, Mohtadi et al. 2016). Moreover, as one consequence of global change, the intensity and
timing of tropical rainfall have been predicted to change dramatically in the upcoming
decades (Lehmann et al. 2015) and the impact that such changes might have on the
reproductive calendar of tropical species remains unknown (Chambers et al. 2013).
It is also uncertain if changes in rainfall could cause a desynchronization between breeding
and seasonal peaks of arthropod prey abundance, as it occurs in temperate latitudes (when
consumer and prey are not reacting to the same cue and/or time window, e.g. Both and Visser
2001, Chambers et al. 2013, Visser et al. 2004, 2006a, Schaper et al. 2012). However, no
conclusions can be drawn about decoupling of breeding phenology and food abundance in the
tropics from extrapolating changes in temperate regions (Stenseth and Mysterud 2002, Hilden
et al. 2012). For instance, it is relatively unexplored how food abundance fluctuates in the
tropics and how females match their breeding calendar to such fluctuations, how females
responsiveness vary and if so, what are the consequences for productivity (but see Grant et al.
2000). Alternatively, the importance of timing might not be so relevant for tropical species,
20
since they are multi-brooded and might be less sensitive to shifts in the timing of climatic
cues (Dunn and Moller 2014).
Here we quantify the environmental triggers of breeding and food phenology of a
resident tropical bird with a flexible breeding calendar. To do so, we use five year data from a
detailed individual-based study on the purple-crowned fairy-wren, an insectivorous passerine
species of the monsoonal tropics (Rowley and Russell 1997, Peters et al. 2013). This is a
sedentary cooperatively breeding bird that depends exclusively on the arthropod prey found in
their year-round territories. First, we use a recently developed systematic approach (van de Pol
et al. 2016) to identify the critical climatic factors and time frame that predict female nesting
activity and arthropod abundance (a proxy for food availability). Subsequently, we examine
the consequences of the identified climatic cues on reproductive output.
Methods
Study species and site
A population of purple-crowned fairy-wrens Malurus coronatus coronatus was monitored at
Australian Wildlife Conservancy’s Mornington Wildlife Sanctuary, located in the monsoonal
tropics of northwest Australia (17°31’ S, 126°6’ E). Birds in our study population occupied
year-round territories along ~15 km of river along Annie Creek and Adcock River (Kingma et
al. 2009) and were individually colour banded and followed throughout their lives. Purple-
crowned fairy-wrens are riparian habitat specialists, and territories are associated with the
evergreen palm-like Pandanus aquaticus. Each territory is defended by a dominant pair,
sometimes accompanied by a group of subordinates which are mostly (~60%) offspring of the
dominant pair; group size ranges from 2 to 10 birds. Only the dominant pair reproduces but
subordinates participate in offspring care, thereby improving reproductive success and breeder
survival (Kingma et al. 2009, 2010). In our population breeding can occur in every month of
21
the year, but there are peaks during and after the wet season (Dec-Mar) and often, but not
always, a peak at the dry season (Sep-Oct) (Rowley and Russell 1993, Peters et al. 2013, Fig
S1).
Nesting
Between July 2006 and November 2010 we recorded 665 nesting attempts (477 found before
eggs, 143 with eggs and 45 with nestlings). Open cup nests are built exclusively by the
dominant female over a period of approximately 10 days prior egg-laying (Hall and Peters
2009). Nest material included flood debris, pieces of Pandanus leaves (36%), dried grass, stems
and rootlets (21%) forming a coarse outer layer lined with a basket of finely woven material
(M. Hall unpublished data). Nest never contained feathers and most nests were built in
Pandanus (96% of 743 nests). Laying date and clutch size (2-4 eggs) were determined by
checking nest contents at three-day intervals after building was complete until clutch
completion. Information on reproductive output was analysed only for nests with known clutch
size. Incubation lasts ca. 12 days (Rowley and Russell 1993).
When nestlings were 7 or 8 days old, they were banded with a unique combination of coloured
bands and one numbered band, weighed and a blood sample taken for parentage analysis and
sex determination (Kingma et al. 2011a). Nestlings usually left the nest 12 or 13 days after
hatching (n = 16 nests; unpubl. data), and after the expected fledging date, we searched the nest
area to locate the fledglings. Territories were visited weekly to confirm the presence of
fledglings, which remain with the family group. To have an estimation of post-fledgling
provisioning time, we conducted 20-minute focal observations of fledglings ranging in age
from 2-16 weeks post-fledging and recorded when fledglings were fed by group members or
foraged independently. The number of fledglings surviving up to 6 weeks after fledging (~9
weeks old) were quantified for successful nests (those not lost due to predation or flooding).
22
Environmental variables
Rainfall and temperature. Our study site receives 89% of the total annual rainfall (average ±
SD = 772 ± 92 mm) during the ‘wet season’ from December to March. There is considerable
inter-annual variation in the onset and intensity of the wet and dry seasons (Brown and Shine
2006, see also Fig. S1). Daily records of local rainfall were obtained by a weather station at our
study site since 2004 (Australian Bureau of Meteorology (BOM) weather station 002076,
Fig.1). Temperatures are generally high – monthly mean maxima range from 28 to 42°C (in
November), and monthly mean minima range from 13 (in June) to 32°C, see Fig. S1). Daily
records of temperature for Mornington Wildlife Sanctuary (our study site) were only available
for 30 months (January 2008-July 2010) and therefore we obtained long-term daily temperature
data from Fitzroy Crossing (BOM weather station 003093, 18.19°S, 125.56 °E, ~ 95 km SW
from our study area) (Fig. S1). Daily temperature at Fitzroy Crossing is strongly correlated
with temperature data obtained at our study site (Spearman Rho = 0.94, p < 0.001, n = 30
months).
Arthropod abundance. This was recorded monthly from September 2007 until June 2010
(Fig.1). The exact diet composition for purple-crowned fairy-wrens is unknown, but all fairy-
wrens are generalist insectivores that eat a wide taxonomic range of small arthropods (Rowley
and Russell 1997). After an initial sampling in all territories, six representative territories were
chosen by their low, medium and high abundance of arthropods (3 sampling sites at each river).
Each sampling was performed at the middle of each month in those representative territories.
In total, 3 sets of 15 sweeps were done in each territory (midpoint and two other points
separated 10 m upstream and downstream; for details see Kingma et al. 2011a). The monthly
23
average of summed length of all arthropods was used as an indicator of food abundance
(Kingma et al. 2011a).
Territory quality. Pandanus cover was quantified as an index of habitat quality of the breeding
territory (for details see Hidalgo Aranzamendi et al. 2016). Birds do not occupy areas without
Pandanus and there is large variation in Pandanus distribution between territories (Kingma et
al. 2011a). Territories with greater Pandanus cover suffer from lower rates of nest predation
(Chapter 2). Moreover, territories with greater Pandanus cover have more subordinates,
indicating that they are more productive or more attractive for subordinates to delay dispersal
(Kingma et al. 2011a) and are preferred by females dispersing after divorce (Chapter 4).
Number of helpers in a territory was considered an additional component of territory quality,
because breeders derive reproductive and survival benefits from larger groups (Kingma et al.
2010, Chapter 2).
Statistical analyses
We analysed the influence of environmental variables on timing of reproduction, female
condition, arthropod abundance and productivity in three steps. First, (A) we identified the
sensitive time window for various climate variables that could affect: (i) nest initiation at the
individual level (start of nest building), (ii) female condition at the individual level (from year-
round captures) and a sensitive time window for (iii) arthropod availability (arthropod
abundance) for the population. Then, (B) once a sensitive window for each climate variable
was identified; we used this time window to find the best supported models that determined
the magnitude of the three biological responses. Third, (C) we determined the effect of the
identified environmental factors on reproductive output of the nesting attempt (clutch size,
24
nestling condition and number of recruits). Explanatory variables included were not strongly
correlated (all |r| < 0.58). All analyses were done in R 3.2.0 (R Core Team 2015).
(A) Identifying the critical climatic time window. We used critical climate window modelling
(van de Pol and Cockburn 2011) to identify the climatic time window that best predicts
biological responses. This approach allows selecting the period in which a biological response
is most strongly affected by climate (Bailey & van de Pol 2015). This method also allows
distinguishing short and long-lag effects of climate signals that might be acting differentially
among individuals when events occur at different times of the year (i.e. relative window, van
de Pol & Cockburn 2011, van de Pol et al. 2016). First we identified which climatic windows
determined (i) nesting activity. Using the software climwin and the slidingwin function (Bailey
& van de Pol 2015), we analysed year-round nesting data using a time-to-event Cox’s
proportional hazard analysis (Cox’s HA) with calendar date of events as a response variable
(nesting = 1, not nesting = 0, grouped per two weeks for each pair as the model failed to
converge with daily values) in relation to daily climate data. All nesting attempts (failed and
successful) of each pair were included. We tested cumulative rainfall (linear and quadratic) and
average daily temperature (linear) as explanatory climatic variables (models failed to converge
with temperature as quadratic variable and similarly when interactions were included). Models
with maximum and minimum daily temperature gave similar results as the average temperature
model, and therefore we show results for average daily temperature only. Interactions between
dependent variables were first tested in simple GLMMs (not with climwin) to avoid
convergence issues when running the slidingwin simulation. Since no interactions were found,
we proceeded to run models without interactions We tested several cut-off points for time
windows (8, 6, 3 and 1 months) and based on those tests, we specified a window up to 75 days
backdating from the event of interest for rainfall and up to 180 days for temperature. We used
25
relative time windows, because the timing of trait expression varies substantially among
individuals (i.e. nests can be initiated year-round; for further information on methods see van
de Pol et al. 2016). We also included the outcome of previous nest (failed or successfully
fledged) and nesting interval in weeks (linear and quadratic) as co-factors, since the likelihood
of re-nesting is dependent on these two variables, possibly in a non-linear way (as found for
other multi-brooded species, see Oppel et al. 2013). Second, we tried to identify the critical
climatic time window that affected (ii) female condition, using 5 years of capture data of all
dominant females, here we used a generalized linear mixed model (GLMM) including female
weight as a response variable in relation to daily climate and with female identity as a random
factor. We specified a window up to 180 days backdating from the capture date. Cumulative
rainfall (linear and quadratic) and average temperature (linear) were tested as explanatory
variables. Finally, we aimed to identify the critical climatic time window that affected (iii)
arthropod abundance. For this, we analysed year-round arthropod abundance data using a
linear model (LM) including monthly arthropod abundance as a response variable in relation
to daily climate data. We specified a window up to 180 days backdating from the sample date
each month. We tested cumulative rainfall (linear and quadratic) and average temperature
(linear) as explanatory climatic variables. Monthly cumulative data for arthropod abundance
was available (Sep 2007- Jun 2010) for the entire study area. Using the plotall function from
the climwin package, we visualized information on all fitted climate windows.
(B) Determining the biological response. Using the bestmodel function, we selected the
strongest climate window model based on AICc values, after the simulation identified the
critical time window for nest initiation, female condition and arthropod abundance (Bailey &
van de Pol 2015). To confirm that our observed results were not product of random chance, we
performed a randomization of the data using 5 repeats (Bailey & van de Pol 2015).
26
(C) Climatic effects. Based on the critical time window and climatic parameters identified as
important for nest initiation, we tested if the rainfall experienced in the month prior to egg-
laying (explanatory variable) predicted quantity and quality of offspring, namely: clutch size,
number of fledglings surviving to independence and condition of nestlings. Rainfall was
included as a linear and quadratic term and month of the year was included as a continuous
linear and quadratic variable (month 1=July) to control for temporal variation in nest success
(Chapter 2). Quadratic terms were posteriorly removed if not significant. Two generalized
linear mixed models (GLMMs) with Poisson distribution were built for clutch size (count
variable, range = 2-4) and number of fledglings that survived until 6 weeks after fledging (count
variable, range = 0-4) as dependent variables. Here we excluded nests that failed due to
predation or other causes (Chapter 2). Other explanatory variables included were: habitat
quality and group size and female identity as random term. Since our dataset includes repeated
measures of individuals’ reproductive output (clutch size and number of fledglings) across
multiple climatic situations, we separated the variance associated with multiple measurements
of the same individual in GLMMs into variance associated with measurements of the same
individual (within-individual variance) or measurements of individuals from the same area
(between-individual variance) (Van de Pol & Wright 2009). All models were first fitted with
generalized linear models (GLM) without random terms to test for overdispersion, and we
proceeded to include random terms in the models detailed above because data was not
overdispersed. To test if the rainfall experienced in the month prior of egg-laying affected
nestling condition, a restricted maximum likelihood model (REML) was built including body
mass of the birds captured at day 7 (continuous variable, range = 5-10 g) as dependent variable.
Explanatory variables were: rainfall and month as linear and quadratic terms (quadratic terms
dropped if not significant), habitat quality, group size, sex and time of the day. Nest identity
was added as a random term to control for pseudo replication. Within and between female
27
effects could not be tested here, due to reduced sample size from several nests with complete
information on nestling condition for the same females. All continuous variables were scaled
and centered. Means (x), estimates (β) and their standard error are presented (±SE). Sample
sizes are shown in tables.
Results
The critical climate window predicting timing of breeding and female condition
Purple-crowned fairy-wrens initiated nests in all months of the year, with seasonal peaks during
the wet season and sometimes during the dry season (Fig. 1). At the population level,
cumulative rainfall in the month (~from 2 to 24 days) prior to nest initiation was the best
supported climatic signal (Table 1), with the best supported models showing a window opening
24 days and closing 2 days before nest initiation (Table S1, Fig. 2), temperature had no effect
on the probability of nest initiation (β = - 0.14±0.09, p = 0.11, Fig. S2). Increases in cumulative
rainfall in the 3.5 weeks (2-24 days) prior to nest initiation increased the fortnightly probability
of nesting (linear effect, Cox’s HA, β = 0.07±0.02, p < 0.001, Table 1), with the effect of
rainfall on nest initiation levelling off with high rainfall (quadratic effect, β = -0.01±0.00, p <
0.001, Table 1).
Individual factors also affected the likelihood of nest initiation. The probability of nest
initiation increased with the time since the previous nesting attempt ended (i.e. re-nesting
interval) (Cox’s HA, β = 0.13±0.04, p < 0.001) and decreased if the previous nest had failed (β
= - 0.41±0.13, p < 0.001) (Table 1). Re-nesting intervals after successful nests were longer and
more variable than intervals following failed nests (re-nest interval after successful nests =
20±11 weeks, after failed nests =5±8 weeks (raw data), t-test, t = 12.70, p < 0.001).
28
No climate window was identified to explain female condition: rainfall (GLMM, linear effect,
β = 0.01±0.01, p = 0.84, quadratic effect β = 0.002±0.001, p = 0.10) and temperature (GLMM,
linear effect, β = 0.05±0.02, p = 0.14).
The critical climate window predicting timing of arthropod abundance
The timing of arthropod abundance depended on slightly longer-term climate parameters than
timing of breeding. The best supported model indicated that the timing of greatest arthropod
abundance depended on rainfall ~2 months previously, with temperature having no effect (Figs
2 and S2). The best supported models show a window opening 70 days and closing 56 days
before the sampling date (Table S1). Greater total rainfall during this critical window (56-70
days before sampling) resulted in increased abundance of arthropods, with an initial strong
linear effect of rainfall (LM, R2 adjusted = 0.41, β = 0.17±0.04, p < 0.001, Table 1, Fig. 2), and
with effect of rainfall on arthropod abundance levelling off with high rainfall values (quadratic
effect β = -0.14±0.01, p < 0.001, Table 1, Fig. 2).The peak in arthropod abundance following
rain appears to last up to two months (Figs 1 and S1).
Effect of rain during critical climate window on reproductive output
Clutch size averaged 2.93 ± 0.05 eggs (range 2-4, median = 3) and increased with the amount
of rainfall of the month (2-24 days) prior to egg-laying (GLMM, Table 2, Fig. 3a). We found
no quadratic effect of rain and this variable was dropped from the model. No other variable
influenced clutch size (Table 2). When partitioning the effect of rainfall on clutch size to effects
within- and between- females, we found a within-female effect of variation in rainfall on clutch
size (n = 96 females, 532 capture records, Δ rainfall β = 0.07±0.03, p = 0.02) indicating that
when females experienced more rain they laid larger clutches. We did not find a between-
female effect of variation in rainfall with clutch size (mean rainfall β = 0.04±0.03, p = 0.20)
29
and other variables also did not influence clutch size (territory quality β = -0.01±0.01, p = 0.87,
group size β = 0.02±0.02, p = 0.34, month of the year β = 0.01±0.03, p = 0.70).
Greater rainfall prior to egg-laying tended to be associated with greater recruitment: nests that
experienced higher rainfall in the month prior to egg-laying tended to produce more fledglings
(GLMM, β = 0.13±0.07, p = 0.07, Table 2, Fig. 3b). Larger groups tended to recruit more
fledglings (β = 0.08±0.05, p = 0.08) but this was not related to habitat quality or month of the
year (Table 2). When partitioning the effect of rainfall on recruitment to effects within- and
between- females, we did not find within- or between-female effect of variation in rainfall on
number of fledglings (GLMM, mean rainfall β = 0.06±0.05, p = 0.27, Δ rainfall β = 0.08±0.05,
p = 0.14). Rainfall of the month prior to egg-laying did not explain nestling condition (Table
2).
Effect of arthropod abundance on reproductive output
Arthropod abundance was correlated with the number of fledglings surviving to six weeks at
the population level (arthropod abundance in the month of fledgling count with number of
recruits in the same month: Spearman Rho = 0.45, p = 0.007, n = 32 months, Fig. 4).
Discussion
We showed that local climate has a major influence on the seasonal timing of breeding in
purple-crowned fairy-wrens and that rainfall, rather than temperature, is a fine-scale predictor
of nesting and optimal food supply. To our knowledge this constitutes one of the few examples
in which rainfall has been demonstrated to link phenology at two trophic levels: peak biomass
of arthropods and breeding phenology of insectivorous birds. Our findings suggest that the
rapid response of females to rainfall – the same climate signal that drives increases in arthropod
abundance- allows individuals to match the timing of breeding to peak food abundance, thereby
increasing production of offspring that need extended parental care.
30
Rainfall triggers female’s response to initiate nesting
It seems that purple-crowned fairy-wren females are directly responding to rainfall (Lloyd
1999, Whitehead and Saalfeld 2000), since they showed a fast response to rainfall, initiating
breeding very soon after the start of rain (2-24 days) (Table 1). To test whether the observed
pattern was due to some females being better at adjusting their response to climate (i.e.
between-individual effects) or if it was a plastic response in which all females tracked the
quantity of rainfall in their decision to breed (Van de Pol & Wright 2009). We showed here a
within-individual effect on clutch size, indicating that females displayed individual variation
in plasticity in the timing of breeding in regards to rainfall. Breeding in response to rain has
been demonstrated for several tropical species (Lack 1954, Grant et al. 2000, Houston 2013).
Such an efficient response in nesting activity is expected when rainfall is unpredictable, and
this is particularly relevant for tropical and sub-tropical species that inhabit arid and semi-arid
habitats (Hau et al. 2004, Senapathi et al. 2011, Houston 2013). Although the monsoonal
tropics have seasonal rain regimes, the onset and timing of rainfall vary widely among years
(Brown and Shine 2006), suggesting that monsoonal species might adjust their breeding to rain
similar as species from arid environments (Hau et al. 2004). In here, females timed their nesting
to the periods with more food abundance for fledglings, which directly affected productivity.
Females use quantity of rainfall as a cue to anticipate food for offspring: possible
physiological mechanisms
Females experiencing more rain during the critical time window laid larger clutches (Fig. 3a),
resulting in more recruits (Fig 3b). However, responses to rainfall were mostly quantitative
than qualitative, since rainfall of the month prior to egg-laying did not influence nestling
condition (Table 2). Such quantitative responses have been also found for species inhabiting
31
arid environments in Africa, America and Australia (Lack 1954, Lloyd 1999, Grant et al. 2000,
Houston 2013). Females seem to use rainfall as a cue to anticipate food for their offspring,
since female condition is not related to rainfall on any temporal scale (no climatic window
could be identified). A response to rainfall not being mediated by female condition initiating
breeding is in agreement with apparent year-round, opportunistic reproduction of female
purple-crowned fairy-wrens. Although we lack records of gonadal size, our year-round records
of nest initiation on some years but not others, suggest that the physiological mechanisms that
birds use to be ready for egg-laying, could be similar to the one of opportunistic breeders in
arid environments. In here, the reproductive system of birds stays in quiescence, without being
completely shut-down during dry periods until rainfall stimulates reproduction (Wikelski et al.
2000, Hau et al. 2004). Also, probability of repeated breeding was not only determined by
rainfall but also by the outcome of the previous nest (successful or failed) and nesting interval
(Table 1), thus long periods without breeding after nests were repeatedly unsuccessful, could
be directly linked to a decline in rainfall. The effect of rain, in combination with individual
determinants on re-nesting, suggests that gonads are never fully regressed. .
Arthropod abundance as key mechanism
Our results show that the peak of arthropod abundance coincides with the period after fledging
(Fig. 5, as Houston 2013). This in contrast with species in temperate latitudes in which the time
of arthropod peak rather matches the nestling period (Visser et al. 2003, 2006b, Both et al.
2009). The timing matches the long period of parental care that fledglings need, as is common
in tropical species (Russell et al. 2000, 2004). Purple-crowned fairy-wren offspring leave the
nest three to four weeks after the start of incubation (Fig. 5) and receive most of their food via
provisioning until they are ca. 5 to 6 weeks old, after that period they are more likely to find
their own food as they age (Fig. S3). In temperate species, the food supplied to nestlings
32
predicts post-fledgling survival (Naef-Daenzer et al. 2016). Here, we have shown that for a
species that requires longer care, it is possible that food supplied to fledglings is key to predict
recruitment and juvenile survival. For species with long periods of fledgling dependence
(Russell et al. 2000, 2004), matching timing of breeding with fledgling’s demand might be
more relevant than timing it with hatching. This is supported by the positive effect of arthropod
abundance on fledgling recruitment (Fig. 4), which – stochastic catastrophes such as nest
predation or inundation excluded - is more strongly affected by post-fledging survival (30% of
fledglings survive), than by nestling survival (90% of nestlings survive).
Conclusion
Knowledge of climatic factors driving breeding phenology for tropical and southern
hemisphere species is limited. Here we show a tropical species that responds rapidly to rainfall,
with no other climatic cue affecting nesting. Although this species lives in a seasonal
monsoonal environment, it seems that it has adaptations more similar to species living in arid
conditions in their patterns of opportunistic breeding. The lack of explanatory power to predict
the direction of shifts in precipitation for the tropics prevents predicting the effects of ongoing
climate change on phenology (Chambers et al. 2013). However, the rapid response to rain; the
fact that food availability responds to the same cue; the small, multiple broods; and ability to
flexibly regulate individual breeding to climate and past breeding success, gives some
confidence that phenological mismatches of the type encountered by insectivore specialists in
the northern hemisphere (Both and Visser 2001, Visser et al. 2006a, 2006b, 2008, Dunn and
Winkler 2010) are less likely to be imminent for such tropical species.
33
Acknowledgments
We thank K. Delhey, K. Tuft, N. Teunissen, M. Fan, M. Roast, A. McQueen and J. Eastwood
for constructive comments on this manuscript. Funding was provided by a Max Planck
Society Minerva Fellowship and an ARC Future Fellowship (FT 110100505) and Monash
University to A.P. and by a Holsworth Wildlife Research Endowment and Stuart Leslie Bird
Research Award to N.H.A. Research was approved by the Australian Bird and Bat Banding
Scheme, the Western Australian Department of Environment and Conservation, the
Australian Wildlife Conservancy and the ethics committees of the School of Biological
Sciences at Monash University and the Max Planck Institute for Ornithology in accordance
with national and international legislation for animal use in research. Our thanks to numerous
volunteers during fieldwork and the support of staff at Australian Wildlife Conservancy’s
Mornington Wildlife Sanctuary.
References
Bailey, L.D. and M. van de Pol. 2015. R package climwin: Climate Window Analysis.
Statistical software. URL: http://cran.r-project.org/web/packages/climwin/
Both, C., M. and M.E. Visser. 2001. Adjustment to climate change is constrained by arrival
date in a long-distance migrant bird. Nature 411:296-298.
Both, C., M., van Asch, R.G. Bijlsma, A. B. van den Burg and M.E. Visser. 2009. Climate
change and unequal phenological changes across four trophic levels: constraints or
adaptations? Journal of Animal Ecology 78:73-83.
Brown, G. P. and R. Shine. 2006. Why do most tropical animals reproduce seasonally?
Testing hypotheses on an Australian snake. Ecology 87:133-143.
Charmantier, A. and P. Gienapp. 2014. Climate change and timing of avian breeding and
migration: evolutionary versus plastic changes. Evolutionary Applications 7:15-28.
34
Chambers, L. E., R. Altwegg, C. Barbraud, P. Barnard, L. J. Beaumont, R. J. M. Crawford, J.
M. Durant, L. Hughes, M. R. Keatley, M. Low, P. C. Morellato, E. S. Poloczanska, V.
Ruoppolo, R. E. T. Vanstreels, E. J. Woehler and A. C. Wolfaardt. 2013. Phenological
Changes in the Southern Hemisphere PLoS ONE 8: e75514.
Crick H.Q.P., Dudley C., Glue D.E. and D.L. Thomson. 1997. UK birds are laying eggs
earlier. Nature 388:526–526.
Dawson, A. 2008. Control of the annual cycle in birds: endocrine constraints and plasticity in
response to ecological variability. Philosophical Transactions of the Royal Society B
363:1621-1633.
Dunn, P. O. and D. W. Winkler. 2010. Effects of climate change on timing of breeding and
reproductive success on birds. In Effects of Climate Change in Birds. Ed Møller et al.
Oxford University Press.
Dunn, P. O. and A. P. Møller. 2014. Changes in breeding phenology and population size of
birds. Journal of Animal Ecology 83:729-739.
Grant, P. R., B. R. Grant, L. F. Keller and K. Petren. 2000. Effects of El Niño events on
Darwin's Finch productivity. Ecology 81:2442-2457.
Hau, M. 2001. Timing of Breeding in Variable Environments: Tropical birds as model
systems. Hormones and Behavior 40: 281-290.
Hau, M., M. Wikelski, H. and J. C. Wingfield. 1998. A neotropical forest bird can measure
the slight changes in tropical photoperiod. Proceedings of the Royal Society B 265:89-95.
Hau, M., M. Wikelski, H. Gwinner and E. Gwinner. 2004. Timing of reproduction in a
Darwin's finch: temporal opportunism under spatial constraints. Oikos 106:489-500.
Hau, M., N. Perfito, I. T. Moore. 2008. Timing of breeding in tropical birds: Mechanisms and
evolutionary implications. Ornitologia Neotropical 19:39-59.
35
Hidalgo Aranzamendi N., M. L. Hall, S. A. Kingma, P. Sunnucks andA.Peters. 2016. Incest
avoidance, extra-pair paternity and territory quality drive divorce in a year-round
territorial bird. Behavioral Ecology 10.1093/beheco/arw101.
Hille, S.M. and C. B. Cooper. 2015. Elevational trends in life histories: revisingthe pace-of-
life framework. Biological Reviews 90: 204–213.
Hindell M. A., C. J. A. Bradshaw, B. W. Brook, D. A. Fordham, K. Kerry, C. Hull and C. R.
McMahon. 2012. Long-term breeding phenology shift in royal penguins. Ecology and
Evolution 2:1563–1571.
Hodges, C. J., E. K. Bowers, C. F. Thompson and S. K. Sakaluk. 2015. Cascading costs of
reproduction in female house wrens induced to lay larger clutches. Journal of
Evolutionary Biology 28:1383-1393.
Houston, W. A. 2013. Breeding cues in a wetland-dependent Australian passerine of the
seasonally wet-dry tropics. Austral Ecology 38:617-626.
Hughes, L. 2000. Biological consequences of global warming: is the signal already apparent?
Trends in Ecology and Evolution 15:56–61.
Kingma, S. A., M. L. Hall and A. Peters. 2011a. No evidence for offspring sex-ratio
adjustment to social or environmental conditions in cooperatively breeding purple-
crowned fairy-wrens. Behavioral Ecology and Sociobiology 65:1203–1213.
Kingma, S. A., M. L. Hall and A. Peters. 2011b. Multiple benefits drive helping behaviour in
a cooperatively breeding bird: An integrated analysis. The American Naturalist 177: 486-
495.
Kingma, S. A., M. L. Hall and A. Peters. 2012. Sperm storage reflects within- and extra-pair
mating opportunities in a cooperatively breeding bird. Behavioral Ecology and
Sociobiology 66:1125-1123.
36
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology 9:10.1186/1472-6785-9-15.
Kingma, S.A., M.L. Hall, E. Arriero & A. Peters. 2010. Multiple benefits of cooperative
breeding in purple-crowned fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79: 757-768.
Lloyd, P. 1999. Rainfall as a breeding stimulus and clutch size determinant in South African
arid-zone birds.Ibis 141:637-643.
Lack, D. 1954. The Natural Regulation of Animal Numbers. Oxford University Press.
Lehmann, J., D. Coumou and K. Frieler. 2015. Increased record-breaking precipitation events
under global warming. Climatic Change 132:501-515
Mohtadi, M., M. Prange and S. Steinke. 2016. Palaeoclimatic insights into forcing and
response of monsoon rainfall. Nature 533:191-199.
Møller A. P., D. Rubolini and E. Lehikoinen. 2008. Populations of migratory bird species that
did not show a phenological response to climate change are declining. PNAS 105:16195–
16200.
Naef-Daenzer, B. and M. U. Gruebler. 2016. Post-fledging survival of altricial birds:
ecological determinants and adaptation. Journal of Field Ornithology 87:227–250.
Oppel, S., G. M. Hilton, R. Allcorn, C. Fenton, A. J. Matthews and D. W. Gibbons. 2013.
The effects of rainfall on different components of seasonal fecundity in a tropical forest
passerine. Ibis 155:464-475.
Pakanen, V., M. Orell, E. Vatka, S. Rytkönen and J. Broggi. 2016. Different ultimate factors
define timing of breeding in two related species. PLoS ONE 11:e0162643
Peters, A., S. A. Kingma and K. Delhey. 2013. Seasonal male plumage as a multi-component
sexual signal: insights and opportunities. Emu 113:232-247.
37
R Core Team. 2015. R: A language and environment for statistical computing. R Foundation
for Statistical Computing. Vienna, Austria. http://www.R-project.org/.
Root, T. L., J. T. Price, K. R. Hall, S. H. Schneider, C. Rosenzweig and J. A. Pounds. 2003.
Fingerprints of global warming on wild animals and plants. Nature 421:57-60.
Rowley, I. and E. Russell. 1997. Fairy-wrens and Grasswrens. Oxford University Press. New
York.
Russell, E. M. 2000. Avian life histories: is extended parental care the southern secret? Emu
100: 377-399
Russell, E. M., Y. Yom-Tov and E. Geffen. 2004. Extended parental care and delayed
dispersal: northern, tropical, and southern passerines compared. Behavioral Ecology 15:
831-838.
Schaper, S. V., A. Dawson, P. J. Sharp, P. Gienapp, S. P. Caro and M. E. Visser. 2012.
Increasing temperature, not mean temperature, is a cue for avian timing of reproduction.
The American Naturalist 179:E55-E69.
Schwartz, M. D. 2003. Phenology: an integrative environmental science. Springer
Shine, R. and G. P. Brown. 2008. Adapting to the unpredictable: reproductive biology of
vertebrates in the Australian wet–dry tropics. Philosophical Transactions: Biological
Sciences 393:363-373
Stenseth N. C. and A. Mysterud. 2002. Climate, changing phenology, and other life history
traits: Nonlinearity and match-mismatch to the environment. PNAS 99:13379-13381.
van de Pol, M., and J. Wright. 2009. A simple method for distinguishing within-versus
between-subject effects using mixed models. Animal Behaviour, 77, 753-758.
van de Pol M. and A. Cockburn. 2011. Identifying the critical climatic time window that
affects trait expression. The American Naturalist 177: 698-707.
38
van de Pol, M., L. D. Bailey, N. McLean, L. Rijsdijk, C. Lawson and L. Brouwer. 2016.
Identifying the best climatic predictors in ecology and evolution. Methods in Ecology
and Evolution doi 10.1111/2041-210X.12590.
Visser, M. E. 2008. Keeping up with a warming world; assessing the rate of adaptation to
climate change. Proceedings of the Royal Society of London Series B 275: 649-659.
Visser, M. E., F. Adriansen, J. H. van Balen, J. Blondel, A. A. Dhondt, S. van Dongen, C. du
Feu, E. V. Ivankina, A. B. Kerimov, J. de Laet, E. Matthysen, R. McCleery, M. Orell and
D. L. Thomson. 2006a. Variable responses to large-scale climate change in European
Parus populations. Proceedings of the Royal Society of London Series B 270:367-372.
Visser, M. E., L.J.M. Holleman and P. Gienapp. 2006b. Shifts in caterpillar biomass
phenology due to climate change and its impact on the breeding biology of an
insectivorous bird. Oecologia 147:167-172.
Visser, M.E., C. Both and M. Lambrechts. 2004. Global climate change leads to mistimed
avian reproduction. Advances in Ecological Research 35:89-110.
Whitehead, P. J. and K. Saalfeld. 2000. Nesting phenology of magpie geese (Anseranas
semipalmata) in monsoonal northern Australia: responses to antecedent rainfall. Journal
of Zoology 251:495-508.
Wikelski, M., M. Hau and J. C. Wingfield. 2000. Seasonality of reproduction in a neotropical
rain forest bird. Ecology 81:2458-2472.
Wikelski, M., L. Spinney, W. Schelsky, A. Scheuerlein and E. Gwinner. 2003. Slow pace of
life in tropical sedentary birds: a common-garden experiment on four stonechat
populations from different latitudes. Proceedings of the Royal Society of London Series
B270:2383–2388
Wingfield, J. C., T. P. Hahn, R. Levin and P. Honey. 1992. Environmental predictability and
control of gonadal cycles in birds. Journal of Experimental Zoology 261:214–231.
39
Table 1. Rainfall in the previous month (2-24 days, linear and quadratic) was the best supported
climatic signal influencing the probability of (a) nest initiation, and rainfall of the 56-70 days
(linear and quadratic) prior to the sampling date was the best supported climatic signal
explaining (b) arthropod abundance. No significant effect of temperature was found on
phenology at either tropic level. The probability of nest initiation was affected also by the time
since the previous nesting attempt (nest interval) and the outcome of the previous nest
(successful or not). Results p < 0.01 shown in bold.
(a)Nesting Predictor Effect SE Z pRainfall (cm) 0.07 0.02 4.02 < 0.001Rainfall (cm)2 -0.01 0.00 -4.09 < 0.001Temperature -0.14 0.09 -1.61 0.11Nest Interval 0.13 0.04 3.13 0.001Nest Interval2 -0.01 0.01 -3.80 0.001Previous nestsuccessful (yes/no) -0.41 0.13 -3.12 0.001
(b)Arthropod abundance Predictor Effect SE Z p
Rainfall (cm) 0.16 0.01 2.09 <0.001Rainfall2 (cm) -0.14 0.01 -2.83 <0.001Temperature 13.81 17.49 0.79 0.44
40
Table 2. Effects of rainfall in the month prior to egg-laying (identified as the critical time window for nest initiation, Table 1) on reproductive
output in purple-crowned fairy-wrens. Dataset includes first and re-nesting attempts. Results of generalized linear mixed models (a-b) with
Poisson distribution and restricted maximum likelihood model (c) with normal distribution. Random factors did not contribute to variance in the
models. Shown are sample sizes, model estimates, their standard errors (SE), z-value and significance, results p < 0.01 indicated in bold.
* non-failed nests only (i.e. excluding nests lost to predation, flooding and other causes of total nest failure).
**body mass on day 7.
*** estimate is male relative to female (males are larger)
Response variable (a) Clutch size (2-4) (b) Number of fledglings (0-4)* (c)Nestling condition (body mass, g)**Number of observations n= 532 nest (171 pairs) n=192 nests (111 pairs) n=357 nestlings (160 nests)Fixed effects Estimate SE z-value p Estimate SE z-value p Estimate SE z-value pRainfall of the monthprior-to egg laying 0.07 0.02 2.63 0.002 0.13 0.07 1.84 0.07 -0.10 0.10 -1.04 0.30Habitat quality -0.01 0.01 -0.12 0.89 -0.01 0.01 -1.02 0.31 -0.01 0.08 -0.15 0.88Group size 0.02 0.02 0.96 0.34 0.09 0.05 1.76 0.08 0.11 0.09 1.26 0.21Month (1=July) 0.01 0.03 0.37 0.71 -0.04 0.09 -0.42 0.67 0.08 0.09 0.85 0.40Time of the day n.a. - - - n.a. - - - -0.01 0.08 -0.11 0.91Sex *** n.a. - - - n.a. - - - 0.25 0.07 3.49 < 0.001
41
Figure 1. The number of territories initiating nests (black columns) is a response that follows cumulative rainfall of the previous month (blue line).
Arthropod abundance follows rainfall with approximately 2 month-lag (green line). The number of fledglings surviving 6 weeks after fledging
(orange column) peaks at the same time as arthropod abundance. Arthropod abundance represented by average of summed length of all arthropods
in mm for Sep 2007 to Jun 2010.
42
Figure 2. Model support for best window (lower Δ AICc) for rainfall in relation to (a) nesting
initiation (2 - 24 days) and (b) arthropod abundance (56 - 70 days). Window open (y-axis)
and window close (x-axis) in days before timing of nest initiation or measurement of
arthropod abundance. Models with the lowest ΔAICc (red and pointed by arrow) are the best
supported and indicate the number of days of a relevant window.
(b)(a)
43
Figure 3. (a) Clutch size and (b) number of recruits (fledglings that survived to 6 weeks after
fledging), increased with the amount of rainfall in the month prior to egg-laying. Dashed line
show regression line through the intercept. Results from generalized linear mixed models with
Poisson distribution. Boxplots’ width adjusted to sample size showing the interquartile range
(box), medians, SE and dots show outliers, n = number of nests
(a)
(b)
Rainfall month prior to egg-laying (mm)
44
Figure 4. Monthly arthropod abundance (average of summed length of all arthropods)
correlates with the number of recruits (number of fledglings that survived to 6 weeks after
fledging) at the population level. Data from September 2007 until June 2010. Dashed line
show regression line through the intercept. Results from a Spearman correlation.
45
Figure 5. Schematic representation of timing of rainfall in relation to nesting initiation and arthropod abundance. Red arrows represent
the positive effect of cumulative rainfall during the critical climate window on clutch size, number of fledglings surviving to 6 weeks
(recruits) and arthropod abundance, the green arrow indicates the direct positive effect of arthropod abundance on recruitment.
46
Supplementary Material
Table S1. Identifying the strongest critical climatic time window that affects (a) nesting and (b)resource phenology (arthropod abundance). Shown five best models for the effects of cumulativerainfall (cm) ordered by ΔAICc with most negative values first. ΔAICc is computed against the nullmodel on no climate effects. The quadratic effect of rainfall was the best supported signal for bothnesting and resource phenology. Information from all fitted climate windows (days before event).Window type is “relative”, range time period is: 0-75 days for nesting and 0-180 days for arthropodabundance*
(a)Nesting ΔAICc Window opens Window closes β SE ω
-40.28 24 2 0.20 0.03 0.08-40.14 29 0 0.17 0.02 0.07-39.65 24 3 0.20 0.04 0.06-39.47 29 1 0.18 0.03 0.05-39.38 29 2 0.20 0.03 0.05
(b)Insectabundance ΔAICc Window opens Window closes β SE ω
-25.57 70 56 0.54 0.18 0.01-24.55 70 54 0.47 0.18 0.007-24.53 70 55 0.48 0.19 0.007-24.51 68 54 0.52 0.18 0.007-24.31 70 53 0.44 0.18 0.007
*for further details on the model structure refer to Bailey & van de Pol (2015), van de Pol &Cockburn2011 and van de Pol et al. (2016)
47
Figure S1. Seasonality of nesting, climatic variables and arthropod abundance.(A) Monthly means ofdaily maximum and minimum temperature for records from Fitzroy Crossing (003093, ~95Km southWest), mean ± SE for Jun2006 to Nov2010. (B) Percentage of territories with breeding activity(initiating nests and/or until fledglings are present for 3 weeks), mean ± SE for Jul2006 to Nov2010.(C) Mean monthly rainfall, mean ± SE for rainfall data at Mornington Wildlife Sanctuary fromJun2006 to Nov2010. (D) Monthly arthropod abundance represented by total length of all arthropodscaptured per sweep in mm, mean ± SE for Sep2007 to Jun2010
48
Figure S2. Model support for all fitted time windows tried for (ΔAICc) temperature in relation to (a)nesting initiation and (b) arthropod abundance. Panel showing window open (y-axis) and windowclose (x-axis) in which models with the lowest ΔAICc (red) are the best supported. Notice lack ofwindow for temperature which has been further confirmed with randomization of the data. Thepatterns evident in panel b indicate the seasonal variation in temperature (notice scale ~300 days).Strongly supported windows are grouped together. For further information refer to van de Pol et al.(2016).
(a) (b)
49
Figure S3. Proportion of focal watches when fledglings were fed (black line) or when they found theirown food (dotted line). The number of fledglings observed in each age category is shown at the bottomof the graph.
50
Declaration for Thesis Chapter 2
Declaration by candidate
In the case of Chapter 2, the nature and extent of my contribution to the work was the
following:
Nature of contribution Extent of contribution (%)
The key ideas, data collection, data analysis, lab work
and writing of the manuscript were my responsibility.
90
The following co-authors contributed to the work. If co-authors are students at Monash
University, the extent of their contribution in percentage terms must be stated:
The undersigned hereby certify that the above declaration correctly reflects the nature and
extent of the candidate’s and co-authors’ contributions to this work.
Candidate’s Signature Date: 15-10-2016
Main Supervisor’s Signature Date: 15-10-2016
Name Nature of contribution
Anne Peters Assistance with writing manuscripts, design of hypotheses, data
collection, interpretation of results, feedback on thesis
Michelle L. Hall Assistance with writing, data collection, interpretation of results.
Sjouke A. Kingma Assistance with writing, data collection, interpretation of results.
Kaspar Delhey Assistance with writing, data analysis, interpretation of results.
51
Chapter 2
Contrasting effects of climate means and extremes on nest failure and
productivity in an endangered tropical bird
Nataly Hidalgo Aranzamendi1, Michelle L. Hall2,3, Sjouke A. Kingma3,4, Kaspar
Delhey1,3 and Anne Peters1,3
1School of Biological Sciences, Monash University, Australia; 2School of BioSciences,
University of Melbourne, Australia; 3Max Planck Institute for Ornithology, Radolfzell,
Germany; 4Behavioural and Physiological Ecology, University of Groningen, The
Netherlands.
*This chapter has been submitted to Functional Ecology
Abstract
Determining effects of climate on reproduction is vital to predict population dynamics and
extinction risks under future climate scenarios. Such studies are especially needed for tropical
species given their susceptibility to environmental change and limited available information.
We studied fine-scale effects of habitat quality and climate means and extremes on nest
success over 5 years (685 nests) in a resident bird of the monsoonal tropics (Malurus
coronatus). High average rainfall increased the likelihood of nest flooding, responsible for
losses of 15% of nests. Nest flooding was less likely in higher-quality habitats (i.e. with more
vegetation cover). Individual adaptive responses, increases in nest height following flooding,
were only partly successful, possibly because nest predation risk favored lower nest height.
Predation caused 57% of nest losses, and was less likely in higher-quality habitat. Predation
increased with increasing mean temperature, but decreased when temperatures were
extremely high, presumably due to temperature-related variation in reptilian predator activity.
52
Overall, nest failure rates were higher with higher average temperature and rainfall and in
lower quality habitat. For successful nests, extremely hot temperatures reduced hatching
success and high average rainfall reduced fledgling survival. Taken together, these results
suggest that ongoing warming, increased rainfall and habitat degradation are likely to reduce
reproductive success in this endangered bird. Such processes are likely to be important across
similar tropical regions.
Key words: heatwave, global warming, precipitation, temperature extremes, climate change,
tropical
Introduction
Global climate change and habitat degradation have been identified as increasing threats to
biodiversity (Foden et al. 2013, IPCC 2013, Maxwell et al. 2016). Forecasting these effects
on species’ reproduction has become an urgent priority to generate relevant information for
the future management of sensitive species (Foden et al. 2013). To date, most studies have
focused on climatic changes in temperate regions (Møller et al. 2010, Pacifici et al. 2015).
However, most of the world’s biodiversity is concentrated at tropical latitudes and little is
known about how tropical species (will) respond to such changes (Charmantier and Gienapp
2014). Moreover, theoretical models and recent evidence suggest that tropical species are
more vulnerable to the effects of climate change than previously assumed (reviewed in
Şekercioğlu et al. 2012).
Although the pervasive nature of global climate change is now widely recognized, the
majority of studies on the ecological impacts of climate change have focused on average
temperatures, and effects of changes in average precipitation are still poorly understood
(Easterling et al. 2000, Lehmann et al. 2015). Moreover, extreme climatic events are
becoming more frequent globally, with increasing record-breaking high temperatures (IPCC
53
2013) and extreme rainfall events (Lehmann et al. 2015) in the last three decades. Such trends
are particularly important for tropical areas, since temperatures are already high, and their
seasonal patterns strongly defined by rainfall (IPCC 2013). Responses to different
components of the same climatic variable (i.e. means versus extremes) do not necessarily
occur in the same direction and should be therefore considered in conjunction (Bailey and
van de Pol 2016).
Reproduction is a critical part of any life cycle, and understanding the differential impact of
means and extremes of climate on different stages of reproduction is central to understand
how climate affects the fitness of organisms, and ultimately their population dynamics
(Senapathi et al. 2011, Pacifici et al. 2015). For example, in birds - one of the most studied
taxa in relation to global change – total nest failure might be caused by extreme climatic
events (e.g. heat waves, droughts, flooding, fires) affecting productivity at the population
level (Grant et al. 2000, Şekercioğlu et al. 2012). Furthermore, birds could be affected at a
finer-scale: embryo development can be impaired at extreme temperatures (Stoleson and
Beissinger 1999), nestling survival and development affected by increasing mean and
extreme temperatures and precipitation (Senapathi et al. 2011, Cunningham et al. 2013) and
adults can alter their parental care (e.g. disrupting nest attendance or chick provisioning)
(Grant et al. 2000, du Plessis et al. 2012). This highlights the importance of identifying
biologically relevant thresholds of climate that potentially affect the likelihood of success,
development and survival of offspring. At the same time, the effects of climatic means and
extremes might be attenuated or exacerbated according to the quality of the habitat in which
individuals are found (Şekercioğlu et al. 2012). Thus, quantifying habitat quality (e.g.
vegetation cover) and variation in climate simultaneously is essential to understand how all
effects combine and affect the final outcome of reproduction, therefore generating relevant
information for management of species (Şekercioğlu et al. 2012).
54
Here we investigate the effects of variation in means and extremes of rainfall and temperature
on nest success of a tropical resident bird, the cooperatively breeding purple-crowned fairy-
wren (Malurus coronatus coronatus). This endangered bird is endemic of north-west
Australia, with distinctly seasonal but highly variable rainfall and high temperatures typical
of the monsoonal tropics (Hughes 2003, Moise et al. 2015). Using five years of detailed nest
monitoring, we analyze the effects of climate and habitat quality on the causes of total nest
failure, and on several stages of reproduction (incubation, hatching-, fledging- and post-
fledging success) and total productivity of nest attempts.
Methods
Study site and population
A population of purple-crowned fairy-wrens was monitored at Mornington Wildlife
Sanctuary- managed by Australian Wildlife Conservancy and located in northwest Australia
(17°31’ S, 126°6’ E) along ~15 km of river along Annie Creek and the Adcock River
(Kingma et al. 2009). At this site, purple-crowned fairy-wren territories occur exclusively in
riparian vegetation containing the palm-like Pandanus aquaticus. Habitat quality; density,
width and height of Pandanus stands, varies throughout the study site. The study area is
located in the monsoonal tropics, with a prominent wet season during December – March. All
birds in our study population (n = 125 in 2005 increasing to 151 in 2010) were individually
color banded and monitored throughout their life. Each territory is defended by a dominant
pair (Hall and Peters 2008), sometimes accompanied by subordinates (range 2-8
subordinates) which are mostly (~60%) offspring of the dominant pair. Only the dominant
pair reproduces, and subordinates participate in offspring care, improving reproductive
success and breeder survival (Kingma et al. 2009, 2010). Although breeding can occur in
55
every month of the year, there is a clear peak during and after the wet season (Dec-Mar) and
sometimes a small peak during the dry season (Sep-Oct) (Peters et al. 2013).
Nest monitoring
Between July 2006 and November 2010, we recorded 685 nesting attempts (details in Table
S1). Nests are open cup and built exclusively by the breeding female (details in Chapter 1).
For each nest, nest location, height above ground level and location coordinates were
recorded. By far the majority of nests (96%) were built in Pandanus, on average 1.5 ± 0.05 m
above the ground (range 0.15-6.50 m, n = 658). Laying date and clutch size (1-4 eggs) were
determined by checking nest contents at three-day intervals after building was complete.
Eggs hatch 13 days after the last egg is laid. Nestlings were banded on day 7 with a
combination of colored leg bands and a numbered band supplied by the Australian Bird and
Bat Banding Scheme. Nestlings usually leave the nest 12 or 13 days after hatching (n=16
nests; unpubl. data) and the territory area was monitored weekly to locate fledglings and
determine their survival. Fledglings are fed by group members until approximately 13 weeks
of age (11 weeks post-fledge), but frequently find their own food from 9 weeks of age; 90%
of fledglings that reach 9 weeks survive until adulthood (unpubl. data). Nests were
considered to have been depredated if a nest known to have contained eggs and/or nestlings
was found empty, often with clear signs of predation (i.e. damaged, predator seen attacking
the nest, or identifiable scats found inside).
Environmental variables
Rainfall and temperature. Daily rainfall records were obtained from a weather station at
Mornington Wildlife Sanctuary since 2005 (Australian Bureau of Meteorology (BoM)
weather station 002076, Fig.S1). On average, rain fell on 68 days per year, with daily
56
rainfalls averaging 12 mm (based on n= 285 days with rain from 2005-2010) and extremes >
48.4 mm (95th quantile from data 2005-2010). Daily temperature records for Mornington
were only available from January 2008-July 2010, so we used temperature data from Fitzroy
Crossing, which had records for our entire study period (BoM weather station 003093,
18.19°S, 125.56 °E, ~ 95 km SW from our study area, Fig. S1), and that is strongly correlated
with data from Mornington Sanctuary (Spearman Rho = 0.94, P < 0.001, n = 30 months). The
average daily maximum temperature was 36.2°C (62% of days during our study period
reached temperatures > 35°C) and extremes were > 42.5°C (95th quantile) and > 40°C (90th
quantile from data 2005-2010).
Territory quality. Purple-crowned fairy-wrens in this region do not occupy habitat without
Pandanus and they depend strongly on this vegetation (Kingma et al. 2011a). Pandanus
cover varies widely between territories, and territories with greater Pandanus cover have
more subordinates (Kingma et al. 2011b), indicating that these territories are more productive
or attractive for subordinates. Moreover, dominant females target territories with greater
Pandanus cover during breeder dispersal (i.e. divorce, Hidalgo Aranzamendi et al. 2016).
Therefore we used total Pandanus cover as an index of territory habitat quality (ranging from
1-20, as in Hidalgo Aranzamendi et al. 2016). Group size was also considered as a
component of territory quality because subordinates increase breeder survival and fledgling
recruitment (Kingma et al. 2010).
Statistical analyses
All analyses were done in R 3.2.0 (R Core Team 2015). We followed a mixed modelling
approach with model selection based on AIC coefficients. Interactions between habitat
quality and climate variables were included in the original models; these were all removed
57
since they were not statistically significant (data not shown). Samples sizes vary for each
analysis due to missing data, and are shown in tables.
Total nest failure. A nest was classified as successful if it resulted in at least one fledgling. A
generalized linear mixed model (GLMM) with binomial distribution was built to assess
factors affecting nest success ignoring the cause of failure (successful=1, unsuccessful=0).
Here, 16 nests that experienced partial predation and were eventually abandoned were coded
as unsuccessful. Additionally, we separately investigated factors affecting nest flooding and
nest predation, the dominant sources of nest failure (included only fledged and flooded or
depredated nests respectively). As explanatory variables we included climatic variables
compiled for the duration of the reproductive attempt: average temperature, exposure to
extreme temperature, average daily rainfall and exposure to extreme rainfall. Since the
duration of nest survival correlates with the number of days a nest is exposed, exposure to
extreme temperature/rainfall was coded as a binomial variable (experienced at least one
extreme day = 1, none = 0). In the nest predation model, we used 40°C (90th quantile) as a
threshold for extreme temperatures, because temperatures > 40°C reduce the foraging activity
of monitor lizards in the monsoonal tropics (Greer 1989) and these are main known predators
of riparian species in this area including purple-crowned fairy-wrens (Doody et al. 2015, for
details see results). Month was included as a continuous linear and quadratic variable (month
1= July), to control for temporal variation in nest success. Average temperature and month as
a quadratic term were highly correlated (r > 0.71), as well as average daily rainfall and
extreme rainfall (r > 0.82). Therefore, four models were built using each variable at a time,
here we present the model with the lowest AIC coefficient (see Supplementary Material for
results of alternative models and further details on model selection). Other environmental
variables included were: habitat quality (Pandanus cover), group size, and nest height.
58
Components of nest success. For nests that did not fail completely (i.e. hatched or fledged at
least one young respectively), we tested if environmental variables affected hatching success
(proportion of eggs that hatch), fledging success (proportion of chicks that fledge) and post-
fledging success (proportion of fledglings that survived until week 9). In each model, we
included the climate experienced during each stage for each nest (from egg-laying to hatching
(nests = 208), from hatching to fledging (nests = 168) and 6 weeks from fledging (nests
=168)). We built glmmPQL models, since they were the best fit to our data, using the cbind
command with (number of surviving, number not surviving) as response variables and the
environmental variables detailed above as independent variables. Models of nest failure
performed better (lower AIC) using average daily rainfall (Table S2) and here we only tested
for average daily rainfall. Total productivity of a nest attempt was defined as the number of
fledglings that survived until 6 weeks after fledgling (~9 weeks old). We tested if the number
of fledglings from successful (not depredated, not flooded, not abandoned) nests (Poisson
distribution, 0-4) was affected by climatic and environmental parameters experienced during
the total nest cycle (i.e. from egg-laying until fledglings survived 9 weeks) using a GLMM.
Nests with a value 0 produced at least one fledgling but none survived to week 9.
Results
Factors affecting total nest failure
The majority of breeding attempts failed (77% of 685 nests did not produce a single
fledgling). Nest inundation (flooding) caused 15% of nests to fail (Fig. 1). Low nest height
increased the likelihood of nest flooding, as did high average rainfall (mean flooded nests =
44.3 mm/day, non-flooded nest = 18.8 mm/day, Table 1a, Fig. 2). Although extreme rainfall
and average rainfall predicted nest failure in the same direction, average rainfall was clearly a
59
better predictor of nest flooding (Table 1a, Table S2). On average, females that lost a nest due
to flooding built the subsequent nest in a higher site (n = 77, paired t = -2.08, mean flooded
nests = 1.38 m, mean re-nest after flooding = 1.71 m, p = 0.04; Fig. 2), but 30% of these
subsequent attempts were also lost to flooding. Overall, successfully fledged nests were
located higher than nests lost to flooding (n = 360 nests, t = 2.54, mean flooded nests = 1.35
m, mean fledged nests = 1.68 m, p = 0.01).
Nest predation was the main cause of nest failure (57% of nests; Fig. 1, Table S1). Most
depredated nests were found torn apart. Direct observations and camera trap recordings of
predation events identified reptiles as the primary predators (details in Table S1). Nests were
less likely to be depredated during periods of lower average temperatures and extreme
rainfall (Table 1b, Fig. 1; average rainfall did not appear to have an effect, Table S1).
Predation was also lower when nests experienced temperatures above 40 °C, and in territories
with higher habitat quality (Table 1b). Higher-placed nests were more likely to be predated
(Table 1b, Fig. 2), although this effect was less strong than the opposite effect for flooding
(Table 1a, Fig. 2). Females that lost a nest due to predation built a subsequent nest at a lower
position, although the difference was not very large and not statistically significant (n = 146,
paired t = 1.38, mean depredated nests = 1.66 m, mean re-nest after predation = 1.49 m, p =
0.16; Fig. 2).
Overall, nests were more likely to be successful (i.e. fledge at least one young) when the
average temperature they experienced was lower, when they received less rain on average,
and when located in territories with higher quality habitat (greater Pandanus cover) (Table
1c).
Factors affecting hatching and fledging success
60
Overall, 23% of clutches produced at least one fledgling (i.e. escaped total nest failure, Table
S1). Hatching success (mean = 0.89 ± 0.24 SE of eggs hatched when total nest failure was
excluded) was greater at higher average temperatures but lower if a nest experienced extreme
temperatures above 42.5°C (Table 2a, Table S2 for results of alternative models, Fig. 1). No
climatic variable affected the proportion of hatchlings that fledged (mean = 0.83 ± 0.26 SE of
hatched eggs fledged, excluding total nest failure) (Table 2b, Fig. 1), but larger groups tended
to have higher fledging success (Table 2b). The proportion of fledglings that survived 6
weeks after fledging (mean = 0.65) was lower when rainfall was higher (Table 2c, Fig.1).
Over the entire nest cycle, no climatic variable influenced the number of surviving fledglings
produced per successful nest (mean =1.3 ± 1.0 SE fledglings per nest were recruited from
breeding attempts that did not completely fail). The only marginally significant effect was
that larger groups tended to have higher recruitment per nest (absolute number of chicks,
Table 2d).
Discussion
Understanding the effects of climate variation on the reproductive cycle of animals has
become a research priority in view of the rapid changes in global climate (Senapathi et al.
2011). To date, such effects have been poorly studied in tropical species (Grant et al. 2000,
Senapathi et al. 2011). Here, we reveal the intricate effects of environmental variation on
different aspects of reproduction in a tropical bird, where nesting success varies as a function
of the direction of the effects exerted by climatic means and extremes (Fig.1).
Rainfall. Rainfall affected nest failure in two ways, high average rainfall increased the
chance of flooding (Fig. 1), and at the same time extreme levels of rainfall reduced the risk of
predation (Fig. 1). Reduction of nest predation rates due to rainfall have been recorded for
61
several bird species from temperate and tropical regions (Morrison and Bolger 2002, Oppel et
al. 2013, Anctil et al. 2014), possibly due to the effects on abundance and activity levels of
most reptilian and avian predators (White 2008, Doody et al. 2015). Surprisingly for a bird
that has peak reproduction during the wet season, and despite the larger effect of nest
predation on nest failure, the overall effect of rain on nest survival was negative (Table 1c).
Additionally, high average rainfall reduced fledgling survival for successful nests (Fig. 1).
Since fledglings completely depend on adults for food (Kingma et al. 2010), rainfall
interfering with provisioning by group members is a likely explanation for a decline in
offspring survival (Grant et al. 2000, Anctil et al. 2014, Öberg et al. 2015).
Phenotypically plastic individual responses could mitigate effects of climate change (van de
Pol et al. 2010). Indeed, we found evidence of plastic, adaptive nest site selection in response
to flooding, albeit with limited success. Higher-placed nests were more successful in general,
and following nest loss due to flooding, females built the subsequent nest in a higher location
(Fig. 2). However, a third of those attempts were lost again to floods. Furthermore, predation
risk favors lower nest heights (Table 1b, Fig. 2, lower height is associated with reduced
predation risk), potentially limiting adaptive increases in nest height. Indeed, predation and
flooding seem to be antagonistic forces upon individual responses on nest site selection.
It is unclear how the overall negative effect of high average rainfall on nest success, both at
the level of likelihood of total nest failure and productivity of successful nests, will work out
in the future. This is mainly because current climate model predictions for expected changes
in average rainfall for the monsoonal tropics have low confidence, although all models and
our long-term data (Fig. S3) show no tendency to increase, thus there appears no imminent
threat (Moise et al. 2015). In contrast, there is high confidence that extreme rainfall events for
the tropics and monsoonal Australia are increasing (Lehmann et al. 2015, Moise et al. 2015,
Fig. S3). Although this might result in lower nest failure rates due to predation, it seems more
62
probable that flooding will become more frequent, as it has been recorded for the past three
decades at our study site (Fig. S2).
Temperature. Variation in temperature also had pervasive effects on reproductive success.
Nest predation probability changed with temperature, with evidence for antagonistic effects:
increasing average temperature increased predation, but very high temperatures (at least one
day > 40°C) decreased it (Table 1b, Fig. 1). This pattern matches the optimal foraging
temperature range of reptilian predators, particularly monitor lizards. These are the main nest
predators of riparian bird species in monsoonal Australia, and their activity increases as it
becomes warmer but it stops when temperatures exceed 40°C (Greer 1989, Doody et al.
2015).
Temperature extremes affected viability of eggs: hatching success was reduced after exposure
to at least one day of extreme temperatures (> 42.5°C, Table 2a, Fig. 1). Passerine embryos
optimally develop between 36-40°C and temperatures above 42.5°C arrest development and
are considered lethal (Stoleson and Beissinger 1999). It appears that females are not fully
capable of compensating for extreme temperatures, and eggs presumably became warmer
than optimal, resulting in hatching failure.
In view of the ongoing increase in average temperature and frequency of heat-waves,
worldwide (IPCC 2013) and in Australia (Moise et al. 2015), including our study site (Fig.
S4), several predictions of their impacts on purple-crowned fairy-wrens can be made. Our
data suggests that predation levels will increase with increasing milder temperatures, but this
could be offset by predation rate decrease due to the increase in days over 40°C. However,
once days with temperature > 42.5°C become common, viability of eggs will be at risk.
Ultimately, the net effect of temperature changes over time will thus depend on the nature of
the changes in the temperature distribution, hampering straightforward predictions of climate
63
change impacts, and this could be similar in other birds of similar ecology and size
(McKechnie et al. 2012).
Habitat. The positive effects of habitat quality (Pandanus cover at the territory level) were
pervasive, including a lower likelihood of flooding (Table 1a, Fig. 1) and lower risk of
predation (Table 1b, Fig. 1), which translated into higher nest survival in better quality
habitat (Table 1c, Fig. 1). The reduced risk of flooding is presumably a result of larger,
denser stands of Pandanus offering nest opportunities higher up the bank, and the lower
predation risk probably results from a greater variety of nest sites offering concealment
against predators (Soanes et al. 2015). Although there were no buffering effects of habitat
quality on climatic effects (as indicated by the lack of significant interactions between
climatic variables and habitat quality), and although habitat quality did not affect fledging
success and fledgling survival, the direct and strong effect of habitat quality on nest success
indicates that habitat preservation is critical for this species, as with many other species
(Şekercioğlu et al. 2012, Öberg et al. 2015).
Conclusions. Our results highlight the variety of effects that climate can have on nesting
success. Some of the effects of means and extremes were opposite (e.g. temperature –
predation), some effects were direct (rain – flooding, temperature – hatching success) and
some indirect (via predator activity or bird behavior), and most effects varied with the stage
of the nesting cycle. Our analysis showed not only that the effect of climate is better
understood when analyzed stage by stage, but also that these varied, opposing effects might
hamper species adaptive responses (e.g. nest height). This complexity, potentially inherent to
other tropical systems with high climate variability (Rubenstein et al. 2016), highlights the
challenge of predicting the effects of long-term climatic change on such populations. This is
particularly relevant since most climate studies have focused on temperate populations and
64
temperature-related consequences (Charmantier and Gienapp 2013). Our understanding of the
potential effects of climate change on the life history of tropical species is still incomplete,
which limits the scope of our forecasts and complicates formulation of ameliorating
strategies.
Acknowledgements
We thank M. van de Pol, K. Tuft, N. Teunissen, M. Fan, M. Roast, A. McQueen and J.
Eastwood for constructive comments on this manuscript. Funding was provided by a Max
Planck Society Minerva Fellowship and an ARC Future Fellowship (FT 110100505) to A.P.,
an ARC DECRA Fellowship (DE120102323) to K.D. and a Holsworth Wildlife Research
Endowment and Stuart Leslie Bird Research Award to N.H.A. Research was approved by the
Australian Bird and Bat Banding Scheme, the Western Australian Department of
Environment and Conservation, the Australian Wildlife Conservancy and the ethics
committees of the School of Biological Sciences at Monash University and the Max Planck
Institute for Ornithology in accordance with national and international legislation for animal
use in research.
References
Anctil, A., A. Franke and J. Bêty. 2014. Heavy rainfall increases nestling mortality of an
arctic top predator: Experimental evidence and long-term trend in peregrine falcons.
Oecologia 174: 1033-1043.
Bailey, L. D. and M. van de Pol. 2016. Tackling extremes: Challenges for ecological and
evolutionary research on extreme climatic events. Journal of Animal Ecology 85: 85-96.
Charmantier, A. and P. Gienapp. 2014. Climate change and timing of avian breeding and
migration: Evolutionary versus plastic changes. Evolutionary Applications 7: 15-28.
65
Cunningham, S. J., R. O. Martin, C. L. Hojem and P. A. R. Hockey. 2013. Temperatures in
excess of critical thresholds threaten nestling growth and survival in a rapidly-warming
arid savanna: A study of common fiscals. PLOS One 8: e74613.
Doody, J. S., R. Soanes, C. M. Castellano, D. Rhind, B. Green, C. R. McHenry and S.
Clullow. 2015. Invasive toads shift predator–prey densities in animal communities by
removing top predators. Ecology 96:2544–2554.
du Plessis, K. L., R. O. Martin, P. A. R. Hockey, S. J. Cunningham and A. R. Ridley. 2012.
The costs of keeping cool in a warming world: Implications of high temperatures for
foraging, thermoregulation and body condition of an arid-zone bird. Global Change
Biology 18: 3063-3070.
Easterling, D. R., G. A. Meehl, C. Parmesan, S. A. Changnon, T. R. Karl and L. O. Mearns.
2000. Climate extremes: Observations, modelling, and impacts. Science 289: 2068-2074.
Foden, W. B. et al. 2013. Identifying the world's most climate change vulnerable species: A
systematic trait-based assessment of all birds, amphibians and corals. PLOS One 8:
e65427.
Grant, P. R., B. R. Grant, L. F. Keller and K. Petren. 2000. Effects of El Niño events on
Darwin's Finch productivity. Ecology 81: 2442-2457.
Greer, A. 1989. Biology and evolution of Australian lizards. Surrey Beatty & Sons PTY
Limited. Australia
Hall, M. L. and A. Peters. 2008. Coordination between the sexes for territorial defence in a
duetting fairy-wren. Animal Behaviour 76:65-73.
Hidalgo Aranzamendi, N., M. L. Hall, S. A. Kingma, P. Sunnucks and A. Peters. 2016. Incest
avoidance, extrapair paternity and territory quality drive divorce in a year-round
territorial bird. Behavioral Ecology doi. 10.1093/beheco/arw101
66
Hughes, L. 2003. Climate change and Australia: Trends, projections and impacts. Austral
Ecology 28: 423-443.
IPCC. 2013. Climate change 2013: The physical science basis. Contribution of Working
Group I to the Fifth Assessment Report of the Intergovernmental Panel on Climate
Change [Stocker, T.F. et al.(eds.)].Cambridge University Press, Cambridge.
Kingma, S. A., M. L. Hall and A. Peters. 2011a. No evidence for offspring sex-ratio
adjustment to social or environmental conditions in cooperatively breeding purple-
crowned fairy-wrens. Behavioral Ecology and Sociobiology 65:1203–1213
Kingma, S. A., M. L. Hall and A. Peters. 2011b. Multiple benefits drive helping behavior in a
cooperatively breeding bird: An integrated analysis. American Naturalist 177:486–495.
Kingma, S. A., M. L. Hall, E. Arriero and A. Peters. 2010. Multiple benefits of cooperative
breeding in purple-crowned fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79: 757-768.
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology 9(15).
Lehmann, J., D. Coumou and K. Frieler. 2015. Increased record-breaking precipitation events
under global warming. Climatic Change 132: 501-515.
Maxwell S. L., R. A. Fuller, T. M. Brooks and J. E. M. Watson. 2016. The ravages of guns,
nets and bulldozers. Nature 536:143-145.
McKechnie, A. E., P. A. R. Hockey and B. O. Wolf. 2012. Feeling the heat: Australian land
birds and climate change. Emu 112:1-7.
Moise, A. et al. 2015. Moonsoonal north cluster report. Climate change in Australia,
Projections for Australia’s natural resource management regions. Ed. Ekstrom, M. et al,
CSIRO and Bureau of Meteorology. Australia.
67
Møller, A. P., W. Fiedler and P. Berthold. 2010. Effects of climate change in birds. Oxford
University Press.
Morrison, S.A. and D.T. Bolger. 2002. Variation in a sparrow’s reproductive success with
rainfall: Food and predator-mediated processes. Oecologia 133:315-324.
Öberg, M., D. Arlt, T. Pärt, A. T. Laugen, S. Eggers and M. Low. 2015. Rainfall during
parental care reduces reproductive and survival components of fitness in a passerine bird.
Ecology and Evolution 5: 345-356.
Oppel, S., G. M. Hilton, R. Allcorn, C. Fenton, A. J. Matthews and D. W. Gibbons. 2013.
The effects of rainfall on different components of seasonal fecundity in a tropical forest
passerine. Ibis 155: 464-475.
Pacifici, M. et al. 2015. Assessing species vulnerability to climate change. Nature Climate
Change 5: 215-224.
Peters, A., S. A. Kingma and K. Delhey. 2013. Seasonal male plumage as a multi-component
sexual signal: insights and opportunities. Emu 113: 232-247.
R Core Team. 2015. R: A language and environment for statistical computing. R Foundation
for Statistical Computing. Vienna, Austria. http://www.R-project.org/.
Rubenstein, D. R., H. Skolnik, A. Berrio, F.A. Champagne, S. Phelps and J. Solomon. 2016.
Sex-specific fitness effects of unpredictable early conditions are associated with DNA
methylation in the avian glucocorticoid receptor. Molecular Ecology 25:1714-1728.
Şekercioğlu, Ç. H., R. B. Primack and J. Wormworth. 2012. The effects of climate change on
tropical birds. Biological Conservation 148: 1-18.
Senapathi, D., M. A. C. Nicoll, C. Teplitsky, C. G. Jones and K. Norris. 2011. Climate
change and the risks associated with delayed breeding in a tropical wild bird population.
Proceedings of the Royal Society B 278: 3184-3190.
68
Soanes, R., A. Peters, K. Delhey and J. S. Doody. 2015. The influence of nest-site choice and
predator sensory cues on nesting success in the Crimson Finch (Neochmia phaeton). Emu
115: 317-325
Stoleson, S. H. and S. R. Beissinger. 1997. Hatching asynchrony, brood reduction, and food
limitation in a neotropical parrot. Ecological Monographs 67: 131-154
Van de Pol, M. et al. 2010. Effects of climate change and variability on population dynamics
in a long- lived shorebird. Ecology 91: 1192-1204.
White, T.C.R. 2008. The role of food, weather and climate in limiting the abundance of
animals. Biological Reviews 83: 227-248.
69
Table 1. Climatic and environmental factors that affect the probability of nest failure in purple-crowned fairy-wrens. Results of GLMMs with
binomial responses using (a) probability of nest not flooding (b) probability of nest not succumbing to predation and (c) overall probability
of nest not completely failing (producing at least one fledgling)a. Random factor (pair identity) did not contribute to variance of the models
(data not shown). Shown are model estimates (continuous variables are scaled), their standard error (s.e.), t value and significance (p).
Results p < 0.05 shown in bold.
a a nest was classified as successful if at least one fledgling was recorded, success = 1, b coded as a factor (yes= experienced at least one day above threshold, estimateshown relative to yes), for temperature days > 42.5°C for nest success model and > 40°C for nest predation model, for extreme rainfall threshold > 48.4 mm, c climaterecorded during the lifetime of a nest from egg-laying until fledging or nest failure, d month (1 = July) in which the nest was initiated, e habitat quality refers toPandanus cover, fgroup size is number of subordinates and adult birds. n.a. = term not fitted in the models
(a) probability of nest flooding(success = 1)
(b) probability of nest predation(success = 1)
(c) overall probability of nestfailure (success = 1)
No. nests (pairs) 234 (114) 381(141) 501(157)
Parameter Estimate s.e.t
valuep Estimate s.e.
tvalue
p Estimate s.e.t
valuep
Intercept -3.02 0.94 -3.22 0.001 -2.11 0.41 -5.16 <0.001 -2.52 0.39 -6.37 <0.001Nest height (m) 0.9 0.31 2.93 <0.01 -0.22 0.12 -1.85 0.06 -0.08 0.11 -0.73 0.47Extreme temperatures(y/n)b 0.52 0.98 0.53 0.6 0.94 0.3 3.12 <0.01 0.52 0.35 1.47 0.14
Average temperature(°C)c -0.43 0.19 -2.34 0.02 -0.18 0.05 -3.67 <0.001 -0.09 0.04 -2.08 0.04
Extreme rainfall (y/n)b n.a. n.a. n.a. n.a. 0.82 0.41 1.99 0.05 n.a. n.a. n.a. n.a.
Average rainfall (mm)c -0.62 0.11 -5.82 <0.001 n.a. n.a. n.a. n.a. -0.14 0.03 -4.07 <0.001
Month (linear)d -0.28 0.22 -1.31 0.19 0.13 0.04 3.07 <0.01 0.08 0.04 1.96 0.05
Habitat qualitye 0.24 0.06 4.23 <0.001 0.06 0.02 2.64 <0.01 0.09 0.02 3.65 <0.001
Group sizef 0.1 0.19 0.55 0.58 0.13 0.09 1.52 0.13 0.11 0.09 1.27 0.21
70
Table 2. Climatic and environmental factors that affect the different components of nest success in purple-crowned fairy-wrens. Dataset includes
first and re-nesting attempts, excludes nests that were completely unsuccessful. Random factor (pair identity) did not contribute to variance of
the models. (a-c) Results of glmmPQL using a proportion of no. birds hatched/fledged/survived 6 weeks after fledging from the total no. of eggs,
total no. birds hatched, total no. birds fledged respectively. (d) Results of GLMM using number of fledglings as response variable. Shown model
estimates (continuous variables are scaled), their standard error (s.e.), t value and significance (p). Results p < 0.05 shown in bold.
(a) Hatching successa (b) Fledging successa (c) Post-fledging successa (d) Total productivity of asuccessful nestb
No. nests (pairs) 208 (109) 168 (97) 168 (97) 188 (109)
Parameter Estimate s.e.t
valuep Estimate s.e.
tvalue
p Estimate s.e. t value p Estimate s.e.t
valuep
Intercept 4.28 0.79 5.42 <0.01 0.10 0.54 0.18 0.86 0.46 0.50 0.93 0.36 0.11 0.22 0.50 0.62
Nest height (m) 0.12 0.18 0.73 0.47 0.03 0.15 0.16 0.87 n.a. - - - n.a. - - -Extremetemperatures (y/n)c -0.82 0.41 -2.02 0.04 -0.26 0.46 -0.57 0.57 0.66 0.39 1.69 0.10 0.14 0.2 0.75 0.46
Averagetemperature (°C) 0.61 0.18 3.41 <0.01 -0.16 0.23 -0.73 0.47 -0.09 0.21 -0.41 0.68 0.09 0.09 0.97 0.33
Daily rainfall(mm)
0.3 0.17 1.75 0.08 0.17 0.17 0.30 0.30 -0.29 0.14 -1.95 0.05 0.03 0.09 0.35 0.73
Month (linear)d -0.19 0.06 -3.28 <0.01 -0.01 0.06 -0.07 0.95 0.23 0.05 4.63 <0.01 0.35 0.08 4.25 <0.001
Habitat qualitye -0.07 0.06 -1.25 0.21 -0.02 0.03 -0.72 0.47 -0.01 0.03 -0.46 0.65 -0.02 0.01 -1.32 0.19
Group sizef -0.16 0.15 -1.08 0.28 0.21 0.11 1.85 0.06 0.11 0.11 0.99 0.32 0.09 0.05 1.83 0.07
a climate recorded during each stage (e.g. from egg-laying to hatching, from hatching to fledging, from fledging until 6 weeks old), b Total productivity asnumber of fledglings that survived six weeks after fledging. Climate recorded since egg-laying until 6 weeks after fledging (total nest cycle), c days >42.5°Ccoded as a factor (yes= experienced at least one day above this temperature, estimate shown relative to yes), d month (1 = July) at start of nest, e habitat qualityrefers to Pandanus vegetation cover, fgroup size is number of subordinates and adult birds. n.a. = term not fitted in the model
71
Figure 1. Schematic representation of climatic and ecological effects on nest success. Sizes
of boxes represent % of eggs with a particular fate (652 nesting attempts). Success rates are
low (11%, green box); the main cause of failure is predation (57%), followed by flooding
(15%). Problems during development (hatching failure, failure to fledge or recruit) make
smaller contributions to overall failure rates. Orange arrows indicate negative effects on
success and green positive effects. Overall failure is predicted to increase with increasing
rainfall and with climate warming.
72
Figure 2. Adaptive phenotypic plasticity in nest height placement. Lower nest height
(relative to substrate in meters) (a) increases risk of flooding but (b) reduces risk of predation
(Table 1, nest success = 1). On average (c) subsequent nests after flooding were located at
higher locations (P = 0.04) and (d) subsequent nests after predation tended to be at lower
locations, although no significant difference was found (P = 0.16). Grey lines connect nesting
attempts of the same individuals. Boxplots showing the interquartile range (box), medians
(connected by bold horizontal line) and SE (error bars).
73
Supplementary material
Statistical analysis
Climate variables were fitted as linear and quadratic terms, and quadratic terms were
removed if not statistically significant. Interactions between habitat quality and climate
variables were included in the original model and then removed due to not being statistically
significant. All combinations of explanatory variables that did not have strong correlations (|r|
< 0.7) were included in the models. When two variables were highly correlated (|r| > 0.7),
they were removed from the original model containing all variables and analysed in separate
models using one of these variables at a time. Models containing month as a linear term
performed better than those without it (based on AIC). Thus, we kept month as an
explanatory variable to control for some unexplained temporal variation in success.
Removing it from our models did not change the effect size or direction of variables
significantly (results not shown). Models were first fitted using generalized linear models
(GLM) without random terms to test for overdispersion. If data were not overdispersed,
generalized linear mixed models (GLMM) were built with lme4 (Bates et al. 2015) including
pair identity as a random term. If data was overdispersed, we used GLMM with penalized
quasi-likelihood (PQL) models, since such models fitted better our data, with the MASS
package (Venables and Ripley 2002). Models were then ranked by AIC values using the
package MuMin (Barton 2014). The models shown in results are the ones with the lowest
value of AIC. These models were selected if they had more than 2 units of difference with the
alternative model, which applied to all cases (Burham and Anderson 2002) (i.e. only one
model selected to be the best one, outputs of alternative models found below). All continuous
variables were scaled, centered and transformed if needed. Means (x), estimates (β) and their
standard error are presented (±SE).
74
Table S1. Outcomes of nesting attempts. The dataset includes numbers of first and re-nestingattempts. Four complete years are shown and 2010 is an incomplete year with estimates from July toNovember only*.
Nest outcome (numbers) 2006/7 2007/8 2008/9 2009/10 2010* Totals
Failed prior to layingUnknown (predated orabandoned)
18 39 22 37 24
Flooded 3 26 17 4 2
Breeder gone 1 2 1 1 1
N total 22 67 40 42 27 198
Failed during incubationUnknown (eggs foundabandoned)
1 5 2 3 1
Flooded 2 22 15 3 1
Predation** 24 56 21 19 18
N total 27 83 38 25 20 193
Failed after hatching
Unknown (nestlings found dead) 1 2 2 1 0
Flooded 2 6 1 2 0
Breeder gone 2 0 2 0 0
Parasitised 4 0 0 1 0
Predation** 24 25 23 15 23
N total 33 33 28 19 23 136
Success
Nests found with eggs, ns or fle 96 160 98 68 68 490
Clutches hatched 69 74 60 38 46 287
Clutches fledged 36 43 32 22 25 158
**Direct observations and camera traps identified nest predators as: Gilbert’s dragon (n=1, Amphibolurus
gilberti), Mitchell's water monitor (n=3, Varanus mitchelli), Merten’s water monitor (n=1, Varanus
mertensi), yellow-spotted monitor (n = 2, Varanus panoptes) and brown-tree snake (n = 1, Boiga
irregularis). A Brush Cuckoo (Cacomantis variolosus) and a Pheasant Coucal (Centropus phasianinus)
were also seen removing nestlings from a nest.
75
Table S2. Outputs of all models showing climatic and environmental factors that affect the probability of total nest failure in purple-crowned fairy-wrens.Dataset includes first and re-nesting attempts. Model with lowest AIC included in Table 1 in the main manuscript. Random factor (pair identity) did notcontribute to variance of the models. Shown are GLMMs with binomial responses (1/0) using (a) probability of nest not flooding (success=1) (b) probabilityof nest not succumbing to predation (success=1) and (c) overall probability of nest not completely failing (producing at least one fledgling) (success=1).Shown are coefficients (continuous variables are scaled) and their standard error (s.e.), significant values in bold (p < 0.05).
Model AIC Intercept Nestheight
Extremetemperature(yes/no)
Averagetemperature
Extremerainfall(yes/no)
Averagerainfall
Month ofthe Year
Monthquadratic
Habitatquality
Groupsize
a 132.8 -3.02±0.94 0.90±0.31 0.52±0.98 -0.43±0.19 NA -0.62±0.11 -0.28±0.22 NA 0.24±0.06 0.10±0.19
135.8 -3.79±1.02 0.81±0.29 -0.20±0.93 NA NA -0.57±0.11 -0.25±0.24 0.15±0.08 0.24±0.06 0.09±0.19
199.7 -4.43±1.13 0.50±0.24 2.97±0.86 NA -0.43±0.50 NA 0.35±0.19 0.45±0.09 0.21±0.05 0.17±0.17
259.5 -0.79±0.59 0.44±0.19 2.83±0.69 -0.42±0.10 -1.39±0.39 NA -0.03±0.09 NA 0.10±0.04 0.11±0.13
b 484.4 -2.11±0.41 -0.22±0.12 0.94±0.30 -0.18±0.05 -0.82±0.41 NA 0.13±0.04 NA 0.06±0.02 0.13±0.09
489.6 -2.06±0.41 -0.19±0.12 0.92±0.31 -0.16±0.05 NA -0.01±0.04 0.13±0.04 NA 0.07±0.02 0.13±0.09
492.3 -2.27±0.49 -0.21±0.12 0.56±0.27 NA 0.93±0.42 NA 0.09±0.04 0.04±0.02 0.05±0.02 0.11±0.09
498.4 -2.04±0.49 -0.18±0.12 0.49±0.27 NA NA -0.01±0.70 0.08±0.04 0.03±0.02 0.05±0.02 0.11±0.09
c 542.7 -2.52±0.39 -0.08±0.11 0.52±0.35 -0.09±0.04 NA -0.14±0.03 0.08±0.04 NA 0.09±0.02 0.11±0.09
546.8 -2.54±0.44 -0.09±0.12 0.24±0.33 NA NA -0.15±0.04 0.05±0.04 0.01±0.02 0.08±0.02 0.11±0.08
573 -2.78±0.44 -0.08±0.11 0.80±0.33 NA 0.05±0.34 NA 0.08±0.04 0.06±0.02 0.08±0.02 0.12±0.08
574.1 -2.21±0.37 -0.08±0.11 0.95±0.35 -0.15±0.05 -0.15±0.33 NA 0.07±0.04 NA 0.07±0.02 0.13±0.08
* NA= term not included in the model.
76
Table S3. Climatic and environmental factors that affect the different components of nest success in purple-crowned fairy-wrens. Shown effect of month asquadratic variable instead of average temperature. Dataset includes first and re-nesting attempts, excluded unsuccessful nests lost completely to flooding orpredation. Random factor (pair identity) did not contribute to variance of the models. (a-c) Results of glmmPQL using a proportion of no. birdshatched/fledged/survived to 6 weeks after fledging from the total no. of eggs, total no. birds hatched, total no. birds fledged respectively. (d) Results ofGLMM using number of fledglings as response variable. Shown are model estimates (continuous variables are scaled), their standard error (s.e.), t value andsignificance (p).
(a) Hatching successa (b) Fledging successa (c) Post-fledging successa (d) Fledgling recruitmentb
No. pairs(nests)
109(208) 97(168) 97(168) 109(188)
Parameter Estimate s.e. t value p Estimate s.e. t value p Estimate s.e.t
valuep
Estimate
s.e.t
valuep
Nest height 0.02 0.17 0.14 0.88 0.03 0.15 0.17 0.86 n.a - - - n.a - -
Extremetemperatures(y/n)c,d
-0.87 0.4 -2.20 0.03 -0.35 0.44 -0.80 0.43 0.48 0.37 1.28 0.20 0.23 0.18 1.28 0.20
Dailyrainfall(mm)d
0.16 0.18 0.92 0.36 0.18 0.19 0.96 0.34 -0.42 0.16 -2.71 <0.01 0.07 0.07 1.02 0.30
Month(linear)e -0.19 0.06 -3.38 <0.01 -0.01 0.07 -0.15 0.88 0.17 0.07 2.36 0.02 0.10 0.02 4.29 <0.001
Month(quadratic)e -0.09 0.02 -4.12 <0.001 0.02 0.03 0.47 0.64 -0.03 0.02 -1.26 0.21 0.09 0.09 0.97 0.33
Habitatqualityf -0.07 0.05 -1.27 0.21 -0.02 0.03 -0.76 0.45 -0.03 0.03 -0.83 0.40 -0.01 0.01 -1.19 0.23
Group sizeg -0.09 0.15 -0.63 0.53 0.2 0.11 1.80 0.07 0.12 0.11 1.12 0.27 0.09 0.05 1.84 0.07a climate recorded during each stage (e.g. from egg-laying to hatching, from hatching to fledging, from fledging until 6 weeks after), b climate recorded sinceegg-laying until 6 weeks after fledging (total nest cycle), c days >42.5°C coded as a factor (yes= experienced at least one day above this temperature, estimateshown relative to zero), d values recorded according to the time frame of each stage considered, e month (1 = July) at start of nest, f habitat quality refers toPandanus cover, ggroup size is number of helpers and adult birds.
77
Figure S1. Values of daily atmospheric temperatures (°C) from Fitzroy Crossing (weather station = 003093, ~95 Km south West of our study site, data fromBureau of Meteorology- BOM) and daily rainfall data (mm) at our study site Mornington Wildlife Sanctuary (weather station = 002076, data from BOM)showing seasonal variation over the year and days that exceeded upper extreme thresholds (>95th quantile). Each point represents daily values from 1st
January 2006 until 31st December 2010 in that period (a) maximum, (b) minimum and (c) average temperature and (d) rainfall. Lines in (a) indicate 42.5°C(95th quantile) and 40°C (90th quantile) and in (d) values above 48.4 mm (95th quantile).
(a) (b)
(d)(c)
78
Figure S2. Total number of days per complete decade with river levels exceeding 95th percentile(above 13.36 m) at Dimond Gorge (part of the Fitzroy river, next to our study site) (BOM,Department of Water site number = 802737). Records since 1960 until 2009.
79
Figure S3. Rainfall records at Fitzroy Crossing (BOM, weather station = 003093, 18.19°S, 125.56 °E,~ 95 km SW from our study area) since 1910 until 2009. (a) Yearly records of average rainfall (mm).(b)Total number of days per complete decade with daily rainfall exceeding 95th percentile for the area(41.56 mm per day)
(a)
(b)
80
Figure S4. Temperature records at Fitzroy Crossing (BOM, weather station = 003093, 18.19°S,125.56 °E, ~ 95 km SW from our study area) since 1997 until 2014. (a) Yearly records of averagetemperature and (b) Total number of days per year exceeding 90th percentile of temperature.
Num
ber
of d
ays
in a
yea
r ex
ceed
ing
90th
quan
tile
(a)
(b)
81
Declaration for Thesis Chapter 3
Declaration by candidate
In the case of Chapter 3, the nature and extent of my contribution to the work was the
following:
Nature of contribution Extent of contribution (%)
The key ideas, data collection, data analysis, lab work
and writing of the manuscript were my responsibility.
90
The following co-authors contributed to the work. If co-authors are students at Monash
University, the extent of their contribution in percentage terms must be stated:
The undersigned hereby certify that the above declaration correctly reflects the nature and
extent of the candidate’s and co-authors’ contributions to this work.
Candidate’s Signature Date: 15-10-2016
Main Supervisor’s Signature Date: 15-10-2016
Name Nature of contribution
Anne Peters Assistance with writing manuscripts, design of hypotheses, data
collection, interpretation of results, feedback on thesis
Michelle L. Hall Assistance with writing, data collection, interpretation of results.
82
Chapter 3
Limited benefits of mate retention for productivity in a tropical
passerine
Abstract
Species that establish permanent pair bonds might be expected to form such relationships
because of benefits associated with long pair bonds, such as improvements in breeding
performance. Conversely, alternative options might be limited, or have costs, particularly in
year-round pair bonded species, in which case no benefits of long pair bonds might be
evident. Potential benefits of long pair bonds have been investigated in few tropical species,
even though the majority of tropical passerines form stable long-term relationships. Here we
investigate potential benefits of long-term pair bonds in a monogamous tropical passerine,
Malurus coronatus. Our data shows only limited benefits of long pair bonds, in which the
likelihood of nest success tended to increase with long relationships. However these effects
were relatively weak, and did not translate into higher annual productivity. We also found
evidence for between-pairs effects: pairs with longer relationships had higher hatching
success, possibly due to death of low quality individuals. Overall, we believe that long-term
pair bonds are the by-product of site fidelity and the lack of opportunities for partner change.
High quality territories were associated with long pair bonds and the benefits that come from
high quality territories are especially important for long-lived species. To detect benefits of
long-term pair-bonds, our sample size possibly needs to incorporate more years of data
overall, to overcome the effects that climate variation and high predation pressure has on
annual reproductive success.
Keywords: monogamy, pair bond investment, breeding experience, survival
83
Introduction
A monogamous pair bond between a male and a female can be an association to breed only
once (i.e. one breeding season) or a bond that continues for several breeding seasons (Cézilly
and Nager 1995, Fowler 1995, Black 2001). This latter form of long-term social monogamy
is particularly common in birds (Black 1996). The main hypothesis to explain the existence
of long-term pair bonds, known as the pair bond investment hypothesis, postulates that
breeding success increases over time when remaining with the same partner (Mills 1973,
Coulson & Thomas 1983, Bradley et al. 1990, Black 2001, Naves et al. 2007). This effect is a
product of improvements in reproduction-related behaviours of a pair, such as sentinel
behaviour, foraging efficiency, choice of nest sites and/or defence against predators (Coulson
1966, Black 2001). Mate retention could also be favoured if finding a new partner is costly
(Ens et al. 1996, Culina et al. 2014). Although divorce is (often) an adaptive decision
(Choudhury 1995, Black 1996, Culina et al. 2014), reproductive success usually decreases
after partner change (Culina et al. 2014). Therefore, it is expected that maintaining long-term
relationships is an adaptive strategy, particularly if a pair remains together year-round
(Fowler 1995).
An alternative explanation for the existence of continuous pair bonds could be that mate
retention is a result of permanent territory occupation, in which case no increase in
reproductive success might be detected over time (territory fidelity hypothesis) (Freed 1987,
Mock and Fujioka 1990). Such a situation might be particularly common in species living in
saturated environments, in which availability of vacancies are rare and site fidelity high
(Llambias et al. 1998).
The benefits of monogamous long-term pair bonds have been extensively studied in long-
lived species living in temperate regions (Black 1996). However, to date, there is limited
research on potential benefits of long-term monogamy in tropical birds. This is despite the
84
fact that year-round territoriality and long pair bonds are common in tropical species (Fowler,
1995, Stutchbury and Morton 2001, Tobias et al. 2011). Moreover, due to their slow life-
histories and relatively long lifespan, effects of long relationships on productivity might be
common for tropical passerines (Hammers et al. 2012).
Individual breeding experience can be an important determinant of productivity of a pair if
individual reproductive performance increases with breeding experience (Forslund and Part
1995, van de Pol et al. 2006, Sanchez-Macouzet et al. 2014). Thus, disentangling effects of
pair duration and breeding experience is the first step to understand if any improvement in
reproduction is related to long relationships (Fowler 1995, Naves et al. 2007). While some
studies have shown that individual breeding experience can matter, it appears that pair bond
duration is more relevant for some species (e.g. Black et al. 2001, van de Pol et al. 2006).
Here we analysed 10 years of data in a tropical cooperatively breeding bird that establishes
long-term pair bonds, the purple-crowned fairy-wren Malurus coronatus. Purple-crowned
fairy-wrens defend their territories year-round and are socially and nearly genetically
monogamous (Kingma, et al. 2009, 2013). Furthermore, partner change is rare, since survival
is high and divorce rates are low (8% annual divorce rate on average, Hidalgo Aranzamendi
et al. 2016). We tested if there is an effect of pair bond duration on reproductive performance
or survival of breeders while taking into account individual breeding experience.
Methods
Study species
Purple-crowned fairy-wrens are cooperative breeders that nest year-round, although there are
two identifiable breeding peaks in the population: one during the wet season (Dec-Mar) and
occasionally one peak at the dry season (Jun-Aug) (Peters et al. 2013). All birds in the study
population were colour-banded for individual identification. Males are more philopatric than
85
females but many breeding females (59% of breeders) recruit from local nests (Hidalgo
Aranzamendi et al. 2016). Only the dominant male and dominant female reproduce. Both
dominants and a variable number of subordinates (0-10) occupy year-round territories that
occur exclusively in riparian vegetation of the palm-like Pandanus aquaticus. Breeders
remain in close proximity to each other most of the time and they use duets for territorial
defence (Hall and Peters 2008, 2009). Divorce rate is low at the population level (8% on
average), in which a third of divorces occur to end incestuous relationships (i.e. between first-
degree relatives, 64% of relationships end in divorce (18 pairs); Hidalgo Aranzamendi et al.
2016). Extra-pair paternity is low among non-incestuous pairs (3.3% population average)
(Kingma et al. 2009, 2013). Age at first reproduction is highly variable: males started
reproducing between 6 months up to 4.2 years (n = 85, mean = 18 months) and females
between 4 months up to 4.5 years (n = 74, mean = 26 months, unpubl. data). Pairs can remain
together up to 8 consecutive years, on average pair bonds last 2.3 years (median = 1.8 years)
and annual survival for breeders in this population is 82 ± 4.6 SD % for males and 78 ± 5.3
SD % for females (Kingma et al. 2009, unpubl.data 2005-2015).
Data collection
Since 2005 until 2010, all reproductive events were followed closely (all nesting attempts
recorded until failure or when fledglings survived to 12 weeks old) starting with 38 territories
that eventually increased to 69 by the end of 2015, as a result of increase in territory density.
From 2010 onwards, two yearly censuses were conducted after the wet season and at the end
of the dry season following the reproductive peaks of the population. During these, all birds
were resighted, and the number and approximate age of fledglings, if present, was recorded.
All fledglings were banded, genotyped and had paternity assigned (details found in Hidalgo
86
Aranzamendi et al. 2016). Birds were called dead if not resighted in two consecutive
censuses. Estimates of pair bond duration using the biannual census method used in 2011-
2015 closely resemble estimates from the weekly census method used between 2005 and
2010 (r = 0.98) (see Hidalgo Aranzamendi et al. 2016 for details).
Reproductive performance was measured for each pair as likelihood of nest success (at least
one chick fledging = yes), clutch size (range 2-4, mean = 2.93±0.05 SD eggs), hatching
success (proportion of eggs that hatched, range = 0.66 to 1, mean = 0.89 ± 0.24 SD of eggs
hatched, excluding total nest failure) and fledging success (proportion of chicks that fledged,
range = 0.25 to 1, mean = 0.83 ± 0.26 SD of hatched eggs fledged, excluding total nest
failure) for each year between 2005-2010. Fledging success could be only determined for
nests in which at least one egg hatched. Annual fledgling production (the number of
fledglings surviving to independence ~9 weeks old, range = 0-6, mean = 1.97±1.21 SD) was
measured for each year from 2005-2015. We also tested annual fledgling production with a
subset of pairs from 2005-2010 to test more accurate estimates of pair bond duration.
Although results were similar, statistical power decreased with this reduced sample and we
show estimates from 10 years of data for annual fledgling production.
Pair bond duration was quantified as the number of consecutive months a male and a female
were together. We retrieved the breeding history of 96 males and 84 females with known pair
bond duration from 2005-2015 (n = 194 pairs; some individuals occurred with multiple
partners) and quantified reproductive performance per year (starting from July 1st). We
assumed that females entering the breeding population (immigrants) had no previous
breeding experience (n = 32). From the total sample, 85 males and 74 females had known
hatch date (excluding immigrant females), the rest of birds were part of the original study
population and were already breeders or subordinates in their initial capture year. For males
and females with known hatch date, we calculated breeding experience as the number of
87
months elapsed since the first reproductive attempt. In this subset of birds, age and breeding
experience (time as a breeder) were highly correlated (r = 0.90), precluding independent
assessment of their effects, we assume that this situation is most likely true for all birds in the
population. We first tried all statistical models using age (in months, as continuous) instead of
breeding experience (in months) and commensurate with the high correlation between age
and breeding experience, results were nearly identical (Table S1 show results with breeding
experience). Because the analysis using breeding experience as a continuous variable
suggested that the effect of breeding experience was evident only in the first year as a
breeder, we coded breeding experience as a factor, i.e. whether or not individuals had
successfully fledged offspring previously. This allowed us to include birds with known
previous reproductive success, but with unknown exact duration of breeding experience and
to extend our sample size. The association between female and male breeding experience
coded in this way was not significant (r < 0.40, x2 = 1.05, p = 0.31). Incestuous pairs were
excluded from all analyses, since they are more likely to divorce and no incestuous
relationships lasted longer than a year (n = 29 pairs, Hidalgo Aranzamendi et al. 2016), plus
incestuous pairs have high rates of hatching failure (Kingma et al. 2013).
Statistical procedure
We tested if there is an effect of pair bond duration on indicators of reproductive performance
using generalized linear mixed models (GLMM), namely (A) probability of nest success
(failure = predation, predation is the major source of failure (57%), other sources of failure
are due to stochasticity of climate variables, Fig. 1, Chapter 2), (B) clutch size, (C) hatching
success, (D) fledging success and (E) annual production of recruits. A relationship between
clutch size and breeding experience has been recorded for other species (Fowler 1995). A
88
correlation between pair bond duration and likelihood of nest success, hatching or fledging
success or annual fledgling production would indicate an improvement in coordination skills
of the pair members in nest defence or improvement in foraging efficiency as a benefit of
mate retention (van de Pol 2006, Lv et al. 2016). Likelihood of nest predation was fitted with
a binomial distribution, in here a successful nest (n = 287) produced at least one fledgling
(success = 1), failed nests (n = 196) were predated (failure = 0). Clutch size (1-4) and annual
fledgling production (0-6, calculated from total nest attempts of the pair in a year) were fitted
with Poisson distributions and hatching success and fledging success (proportion
hatched/fledged, non-hatched/non-fledged) were fitted with binomial distributions using the
command cbind and lme4 (Bates 2014).
Pair bond duration (months) was entered as a linear and quadratic factor to test for non-linear
effects of pair bond duration on performance (i.e. a quadratic effect could show a decline at
old ages). To disentangle within-pair and between-pair effects of pair bond duration, we
added maximum pair bond duration as a covariate in a separate model; this allows to
distinguish if effects of pair bond duration are due to selective disappearance of particular
pairs, e.g. with low productivity, from the population or due to changes in pair productivity
with changing pair bond duration (van de Pol et al. 2006, van de Pol 2006). Models were first
tested with GLMs with quasi-binomial distributions to account for overdispersion. If models
were not overdispersed we proceeded with GLMMs including the random terms. Other
curvilinear function (logarithmic) of pair bond duration were tried instead of both linear and
quadratic factors (van de Pol 2006), to test if reproductive performance reflects a non-linear
response (e.g. increase in reproductive performance reaches a plateau after some years
together). This function was not significant in any model (results not shown). Since repeated
measurements of the same pair were taken and individuals were also found in different pair
combinations, we included male and female identity and year as cross-classified random
89
effects with individual measurements nested within these random effects (for details van de
Pol and Verhulst 2006, van de Pol et al. 2006).
Because territories with more vegetation cover and larger groups have higher productivity
(Kingma et al. 2011a, Chapter 2), both variables were included in all models and models
were also tested using only these two variables as a baseline model (i.e. excluding effects of
pair bond duration or breeding experience) and then outputs of alternative models were
compared to baseline model. We also included breeding experience of the breeders (as factor
1/0, in which 1 = fledged chicks in a prior year). Correlations between co-variates were all
low (|r| < 0.54). We present models based on AIC coefficients (top models are those < 2 units
of difference from the null model).
We also tested whether long term pair bonds were associated with territory quality (an index
of Pandanus vegetation cover 1-20, Chapter 2), using a GLMM, we entered one value for
territory quality for each pair, and their maximum pair bond duration was included as a linear
and quadratic factor; also initial group size as a covariate. Since males and females were
found in different pair combinations, they were included as random terms. Finally, we tested
the effect of pair bond duration (months) on annual probability of survival of breeders. For
this, we included the last year for each pair with known pair bond duration (pair bonds could
end due to divorce, death of focal bird, death of mate, or be ongoing) and quantified
individual survival probability over the following year. Survival to 30th June of the next year
(x = initial year starting 1st July, x+1 = year finishing at 30th June) was included as a
dependent variable in a GLM with binomial distribution. We included pair bond duration
(months) on 1st July of year x as a linear and quadratic term, fledgling production in last year
x-(x+1), group size and territory quality (at initial year starting on 1stJuly = year x) as fixed
factors. We did not include maximum pair bond duration due to collinearity with age and
breeding experience. Males and females were analysed separately, and since survival can
90
depend on age, we included age as a covariate and analysed birds of known age only (see
Table S2). We repeated these models using breeding experience as a factor and with the
extended sample size to include birds with unknown hatch date, obtaining similar results.
Continuous variables from all models were scaled. All analyses were done in R 3.1.2 (R
Development Team 2016).
Results
Models without pair bond duration and breeding experience showed the lowest AIC
coefficients in most cases (Tables 1 and 2), suggesting that pair bond duration and breeding
experience or age of breeders do not increase reproductive output over time. We detail here
outputs of models that included both terms and specific cases in which some level of
significance was found. The majority of nest attempts failed mostly due to predation (57% of
nests, see also Chapter 2). The likelihood of nest predation tended to decrease with increasing
pair bond duration (positive linear effect, p = 0.09), however it increased again for very long
pair bonds (i.e. negative quadratic effect, p = 0.05, Fig. 1). Breeding experience of the pair
did not affect likelihood of nest predation (Table 1) but larger groups had higher probability
of nest success (Table 1). We identified no between-pairs effect on likelihood of nest
predation (no effect of maximum pair bond duration), meaning that pairs with longer pair
bonds did not systematically differ in nest predation rates.
Pair bond duration did not affect clutch size, hatching success or fledging success (Table 1).
Although hatching success did not change as pairs stayed together longer (Table 1), we did
find a positive between-pairs effect on hatching success (i.e. positive effect of maximum pair
bond duration, p = 0.04) indicating that pairs with higher hatching success stayed together for
longer (Table 1). Annual fledgling production did not vary with pair bond duration (no linear
or quadratic effect, Table 2), or with breeding experience (Table 2). However, territories of
91
higher quality tended to recruit more fledglings (Table 2, p = 0.08). The best supported
models did not include breeding experience, for most reproductive indicators (Tables 1 and
2).
We found a trend for territories of higher quality being held by pairs with longer relationships
(positive linear effect, p = 0.07, Table 3) but this decreased for pairs that stayed together for
very long (i.e. negative quadratic effect, p = 0.03, Table 3, Fig. 2). Similarly, territories of
higher quality had larger groups (Table 3, p = 0.02).
Longer pair bonds were not associated with higher likelihood of surviving the following year
(Table 4), although there was a trend for a negative linear (p = 0.07) and positive quadratic (p
= 0.08) effect of pair bond duration on male survival (Table 4). Also, we found a trend for
lower survival for experienced females (p = 0.06, Table 4).
Discussion
We found limited evidence for positive effects of long pair bonds on productivity in purple-
crowned fairy-wrens. These benefits occurred only at early stages of reproduction, in which
the likelihood of nest success tended to increase over the first years of relationship and then
decrease with longer pair bonds. Increased likelihood of nest success with longer pair bonds
could be related to improvements in nest placement by females. However, this does not seem
particularly likely, because (1) no effect of individual breeding experience was found and (2)
although nest predation risk might be slightly reduced by nest placement at lower heights,
this effect is small and countervailed by increased risk of nest flooding (Chapter 2).
Alternatively, pairs breeding together for several years could improve pair compatibility in
defending nests against predators, something that has been demonstrated for other species
(Black 2001, Griggio and Hoi 2011, Weiß et al. 2010, Leu et al. 2015). Nest predation is
92
responsible for the majority of nests losses in purple-crowned fairy-wrens and successful nest
defence would be a key benefit for pairs that stay together for a long time. However, this
increase in nest survival was moderate and detectable over the first years of relationship only
(Fig. 1). Nest survival declined again for longer pair bonds, which might be related to age,
and an indication of senescence (Fig. 1, Table 1). However, sample sizes for very long pair
bonds are reduced and this decline might be driven by relatively few pairs (Fig. 1).
Pairs that remained together longer had higher hatching success (Table 1), as is demonstrated
by the positive between-pair effect of pair bond duration (maximum pair bond duration) on
hatching success. This relationship suggests selective disappearance of low quality pairs
(pairs with at least one low-quality member), since pairs with short pair bond duration are
disproportionally the result of partner’s death (Chapter 4). Although, we found no strong
effects of pair bond duration on adult survival (Table 4), males tended to survive less well in
short to mid-range pair bonds, with a trend for improved survival at long pair bonds
(quadratic effect). Although there is evidence that long-term pair bonds could involve
survival benefits in other species (Stutchbury and Morton 2001), annual survival in purple-
crowned fairy-wrens is high (up to 81% on average for both sexes, Kingma et al. 2009),
which might hinder detection of survival benefits of long pair bonds.
Alternative benefits, rather than increased productivity or survival, explain the existence of
long-pair bonds, and our results are in agreement with the territory fidelity hypothesis, Freed
1987, Mock and Fujioka 1990, Tobias et al. 2016). This hypothesis suggests that long-term
pair bonds effectively are a by-product of maintaining or permanently using the same site or
territory (Ens et al. 1995, Morton 1996, Black 2001, Gill et al. 2007, Tobias et al. 2016). This
could well apply to purple-crowned fairy-wrens, since pairs that stayed together longer
tended to occupy territories of higher quality (Table 3). In this species, breeding pairs occupy
the same territory for the duration of their pair bond and pairs switching territories together
93
has never been recorded. As is common in cooperative breeders (Komdeur 1996), purple-
crowned fairy-wrens also experience strong constraints on available territories (breeding
vacancies) which is evidenced by delayed breeding and the formation of long-term, up to one
year, incestuous pair bonds, despite significant costs of inbreeding (Kingma et al. 2009, 2013,
Hidalgo Aranzamendi et al. 2016). Also close pair coordination is important in territorial
defence (Hall and Peters 2009). Moreover, in purple-crowned fairy-wrens, high quality
territories are associated with many long-term benefits: high quality territories decrease the
likelihood of nest predation (Chapter 2), drive breeder dispersal in divorcing females
(Chapter 4), tend to recruit more fledglings and hold larger groups (Kingma et al. 2010,
2011b, Table 2) indicating that these could be attractive for subordinates to delay dispersal or
disperse to. All this could explain why pairs that held higher quality territories stayed
together for longer. However, we found that pairs with very long pair bonds hold lower
quality territories, as evidenced by the negative quadratic relationship between pair bond
duration and territory quality (i.e. quadratic effect, Table 3, Fig. 2). This quadratic
relationship could be a reflection that pairs with very long relationships stayed together due to
constraints on availability of any alternative territories, showing that owning a territory is
constrained by opposite forces. Although most of our results offer support for the territory
fidelity hypothesis among all the other hypotheses tested, this option also has less clear
predictions since individual quality or survival (and thereby pair bond duration) could be
interlaced with territory quality. Furthermore, territory quality and productivity of the
territory, as there is a positive relationship between productivity and group size (Kingma et
al. 2010) and between territory quality and productivity (Table 2) are inherently linked and
thus assumptions on which is the reason for long term pair bonds based on those two
indicators might be circular.
94
Although it is assumed that inexperienced breeders are less effective defending territories or
achieving successful reproduction (Grigio and Hoi 2011, Ihle et al. 2015, Lv et al. 2016), we
did not find an effect of breeding experience on any aspect of reproductive success and in
fact, models were better supported when these parameters were not included (Table 1). This
could be because in species living in cooperative societies, individual effects of breeding
experience might be less important or harder to detect (DuVal 2011). In such species,
breeding experience can be indirectly gained by helping, in this situation, many of the skills
required to improve reproductive success, such as sentinel behaviour and better defence
against predators might be gained before acquiring their own offspring while helping raising
younger group members (DuVal 2013).
In summary, we found that mate retention in purple-crowned fairy-wrens conferred limited
reproductive benefits over time. We found limited evidence to support the hypothesis that
long-term pair bonds are maintained to increase productivity. Two other studies of tropical
birds similarly reported no apparent benefits of mate retention for reproduction (House wren
Freed 1987, Pinyon jays Marzluff and Balda 1988), suggesting a lack of benefits of mate
retention on productivity for tropical species more generally. Instead, it seems that the need
for dual defence of high quality territories could be a priority over reproductive benefits.
References
Bates, D., M. Maechler, B. Bolker and S. Walker. 2014. lme4: Linear mixed-effects models
using Eigen and S4. R package version, 1(7).
Black, J. M. 1996. Partnerships in Birds: the Study of Monogamy. Oxford: Oxford University
Press.
Black, J. M. 2001. Fitness consequences of long-term pair bonds in barnacle geese:
Monogamy in the extreme. Behavioral Ecology 12:640-645.
95
Bradley, J. S., R. D. Wooller, I. J. Skira and D. L. Serventy. 1990. The influence of mate
retention and divorce upon reproductive success in short-tailed shearwaters Puffinus
tenuirostris. Journal of Animal Ecology 487-496.
Cézilly, F. and R. Nager. 1996. Age and breeding performance in monogamous birds: The
influence of pair stability. Trends in Ecology & Evolution 11, 27.
Choudhury S. 1995. Divorce in birds: a review of the hypotheses. Animal Behaviour 50:413-
429.
Coulson, J. C. 1966. The influence of the pair-bond and age on the breeding biology of the
kittiwake gull Rissa tridactyla. The Journal of Animal Ecology 269-279.
Coulson, J. C. and C. Thomas. 1983. Mate choice in the kittiwake gull. In: Mate Choice (Ed.
by P. Bateson), pp. 361-376. Cambridge: Cambridge University Press.
Culina, A., R. Radersma and B.C. Sheldon. 2014. Trading up: The fitness consequences of
divorce in monogamous birds. Biological Reviews 90:1015-1034.
DuVal, E. H. 2011. Variation in annual and lifetime reproductive success of lance-tailed
manakins: Alpha experience mitigates effects of senescence on siring
success. Proceedings of the Royal Society of London B: Biological Sciences
rspb20111840.
DuVal, E. H. 2013. Does cooperation increase helpers' later success as breeders? A test of the
skills hypothesis in the cooperatively displaying lance‐tailed manakin. Journal of Animal
Ecology, 82:884-893.
Ens, B. J., Choudhury, S. and Black J. M. 1996. Mate fidelity and divorce in birds. In: Black,
J. M. (ed.). Partnerships in birds. The study of monogamy, Oxford University Press,
Oxford, 344-401.
Forslund, P. and T. Pärt. 1995. Age and reproduction in birds—hypotheses and tests. Trends
in Ecology & Evolution 10: 374-378.
96
Fowler, G. S. 1995. Stages of age-related reproductive success in birds: Simultaneous effects
of age, pair-bond duration and reproductive experience. American Zoologist 35: 318-328.
Freed, L. A. 1987. The long-term pair bond of tropical House Wrens: Advantage or
constraint? The American Naturalist 507-525.
Griggio, M. and H. Hoi. 2011. An experiment on the function of the long-term pair bond
period in the socially monogamous bearded reedling. Animal Behaviour 82: 1329-1335.
Gill, S.A., E.D. Alfson and M. Hau. 2007. Context matters: Female aggression and
testosterone in a year-round territorial neotropical songbird (Thryothorus leucotis).
Proceedings of the Royal Society of London B: Biological Sciences 274: 2187–2194.
Hall, M. L. and A. Peters. 2008. Coordination between the sexes for territorial defence in a
duetting fairy-wren. Animal Behaviour 76: 65-73.
Hall, M. L. and A. Peters. 2009. Do male paternity guards ensure female fidelity in a duetting
fairy-wren? Behavioral Ecology 20: 222-228.
Hammers, M., D. S. Richardson, T. Burke and J. Komdeur. 2012. Age-dependent terminal
declines in reproductive output in a wild bird. PloS One 7: e40413.
Hammers, M., S. A. Kingma, K. Bebbington, J. van de Crommenacker, L. G. Spurgin, D. S.
Richardson, T. Burke, H. L. Dugdale, J. Komdeur. 2015. Senescence in the wild: Insights
from a long-term study on Seychelles warblers. Experimental Gerontology 71: 69–79.
Hatchwell, B. J., A. F. Russell, A. D. C. MacColl, D. J. Ross, M. K. Fowlie, and A.
McGowan. 2004. Helpers increase long-term but not short-term productivity in
cooperatively breeding long-tailed tits. Behavioral Ecology 15:1-10
Hidalgo Aranzamendi, N., M. L. Hall, S. A. Kingma, P. Sunnucks and A. Peters. 2016. Incest
avoidance, extrapair paternity, and territory quality drive divorce in a year-round
territorial bird. Behavioral Ecology arw101.
97
Ihle, M., B. Kempenaers and W. Forstmeier. 2015. Fitness benefits of mate choice for
compatibility in a socially monogamous species. PLoS Biology 13: e1002248.
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology 9:15, 10.1186/1472-6785-9-15.
Kingma, S. A., M. L. Hall, E. Arriero and A. Peters. 2010. Multiple benefits of cooperative
breeding in Purple-crowned Fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79:757-768.
Kingma, S. A., M.L. Hall and A. Peters. 2011a. No evidence for offspring sex-ratio
adjustment to social or environmental conditions in cooperatively breeding purple-
crowned fairy-wrens. Behavioral Ecology and Sociobiology 65:1203-1213.
Kingma, S. A., M. L. Hall and A. Peters. 2011b. Multiple benefits drive helping behaviour in
a cooperatively breeding bird: An integrated analysis. The American Naturalist 177:486-
495.
Kingma, S. A., M. L. Hall and A. Peters. 2013. Breeding synchronization facilitates extra-
pair mating for inbreeding avoidance. Behavioral Ecology doi: 10.1093/beheco/art078.
Komdeur, J. 1996. Influence of age on reproductive performance in the Seychelles
warbler. Behavioral Ecology 7:417-425.
Llambías, P. E., Wrege, P., & Winkler, D. W. (2008). Effects of site fidelity and breeding
performance on mate retention in a short‐lived passerine, the tree swallow Tachycineta
bicolor. Journal of Avian Biology 39:493-499.
Lv, L., J. Komdeur, J. Li, I. B. Scheiber and Z. Zhang. 2016. Breeding experience, but not
mate retention, determines the breeding performance in a passerine bird. Behavioral
Ecology arw046.
98
Marzluff, J. M. and R. P. Balda. 1988. Pairing patterns and fitness in a free-ranging
population of Pinyon Jays: What do they reveal about mate choice? The Condor 201-
213.
Mills, J. A. 1973. The influence of age and pair-bond on the breeding biology of the red-
billed gull Larus novaehollandiae scopulinus. Journal of Animal Ecology 42: 147-162.
Mock, D. W. and M. Fujioka. 1990. Monogamy and long-term pair bonding in
vertebrates. Trends in Ecology & Evolution 5: 39-43.
Morton, E.S. 1996. A comparison of vocal behavior among tropical and temperate passerine
birds. In: Ecology and Evolution of Acoustic Communication in Birds, (D.E. Kroodsma
& E.H. Miller, eds), pp. 258–268. Cornell University Press, Ithaca, New York.
Naves, L. C., E. Cam and J. Y. Monnat. 2007. Pair duration, breeding success and divorce in
a long-lived seabird: benefits of mate familiarity? Animal Behaviour 73: 433-444
Peters, A., S. A. Kingma and K. Delhey. 2013. Seasonal male plumage as a multi-component
sexual signal: insights and opportunities. Emu 113: 232-247.
Pradel, R., R. Choquet and A. Bechet. 2012. Breeding experience might be a major
determinant of breeding probability in long-lived species: The case of the Greater
Flamingo. PLOS One 7:e51016
Pyle, P., W. J. Sydeman and M. Hester. 2001. Effects of age, breeding experience, mate
fidelity and site fidelity on breeding performance in a declining population of Cassin’s
auklets. Journal of Animal Ecology 70: 1088-1097
R Core Team. 2015. R: A language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. http://www.R-project.org/.
Sánchez-Macouzet, O., C. Rodriguez, H. Drummond. 2014. Better stay together: Pair bond
duration increases individual fitness independent of age-related variation. Proceedings of
the Royal Society of London B: Biological Sciences 281: doi: 101098/rspb20132843.
99
Song, Z., and M. W. Feldman. 2013. The coevolution of long‐term pair bonds and
cooperation. Journal of Evolutionary Biology 26: 963-970.
Stutchbury, B. J. and E. S. Morton. 2001. Behavioral ecology of tropical birds. Academic
Press.
Tobias, J. A., V. Gamarra-Toledo, D. García-Olaechea, P. C. Pulgarin and N. Seddon. 2011.
Year‐round resource defence and the evolution of male and female song in suboscine
birds: social armaments are mutual ornaments. Journal of Evolutionary Biology 24:
2118-2138.
Tobias, J. A., C. Sheard, N. Seddon, A. Meade, A. J. Cotton, S. Nakagawa and E. Datasets.
2016. Territoriality, social bonds, and the evolution of communal signaling in
birds. Frontiers in Ecology and Evolution 4: 74. doi: 10.3389/fevo.
van de Pol, M. 2006. State-dependent life-history strategies. A long-term study on
oystercatchers. PhD Thesis. University of Groningen.
van de Pol M., D. Heg, L.W. Bruinzeel, B. Kuijper and S. Verhulst. 2006. Experimental
evidence for a causal effect of pair-bond duration on reproductive performance in
oystercatchers (Haematopus ostralegus) Behavioral Ecology 17: 982-991.
van de Pol, M. and S. Verhulst. 2006. Age‐dependent traits: a new statistical model to
separate within‐and between‐individual effects. The American Naturalist 167: 766-773.
Weiß, B. M., K. Kotrschal, E. Möstl and K. Hirschenhausera. 2010. Social and life history
correlates of hormonal partner compatibility in greylag geese (Anser anser) Behavioral
Ecology 21: 138-143.
100
Table 1. Effect of pair bond duration (PBD), in months and individual breeding experience (BE, as factor+) on productivity. (A) Probability of nest success
(predation = 0, success = 1). (B) Clutch size, (C) hatching success (proportion eggs hatching) and (D) fledging success (proportion chicks fledging).
Reproductive estimates from 2005-2010. n = # nests, F = # females, M = #males, Y = # years. Random factors (female and male identity and year) did
not contribute to variation in any model. Models arranged by AIC values; null models (without pair bond duration and breeding experience) had higher
support, top models (< 2 AIC) are those compared in relation to null model, p < 0.05 in bold.
(A) Nest success n135, 50 F, 50 M, 6YAIC 192.3 192.10 193.70 195.40 196.90 198.20Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 0.12 0.17 0.46 0.01 0.19 0.94 0.01 0.19 0.94 0.47 0.57 0.41 0.48 0.58 0.41 0.13 0.49 0.78Territory quality 0.01 0.18 0.97 0.12 0.20 0.55 0.08 0.21 0.70 0.12 0.21 0.55 0.08 0.22 0.71 0.08 0.19 0.69Group size 0.35 0.18 0.05 0.57 0.24 0.02 0.58 0.24 0.02 0.60 0.25 0.01 0.61 0.25 0.01 0.52 0.22 0.02PBD 1.06 0.63 0.09 1.15 0.65 0.08 1.52 0.85 0.07 1.61 0.87 0.06 -0.17 0.25 0.48PBD ^2 -1.22 0.62 0.05 -1.22 0.62 0.05 -1.54 0.74 0.04 -1.55 0.75 0.04PBD (max) 0.16 0.24 0.50 0.17 0.24 0.49BE F (y/n) 0.27 0.51 0.59 0.29 0.51 0.57 0.05 0.48 0.91BE M (y/n) 0.38 0.56 0.50 0.37 0.56 0.51 0.15 0.48 0.75
(B) Clutch size n187, 55F, 53M, 6YAIC 681.1 685.00 685.40 685.60 687.60 687.80Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 1.27 0.04 <0.001 1.27 0.04 <0.001 1.26 0.04 <0.001 1.35 0.11 <0.001 1.36 0.12<0.001 1.36 0.12 <0.001Territory quality 0.03 0.04 0.52 0.03 0.04 0.52 0.01 0.04 0.80 0.02 0.04 0.58 0.02 0.04 0.60 0.01 0.04 0.90Group size 0.03 0.04 0.41 0.03 0.04 0.46 0.03 0.04 0.46 0.04 0.04 0.38 0.04 0.04 0.37 0.04 0.04 0.37PBD -0.02 0.13 0.86 -0.06 0.13 0.63 0.02 0.05 0.66 0.05 0.16 0.75 0.01 0.17 0.95PBD ^2 0.02 0.12 0.85 0.03 0.12 0.80 -0.03 0.14 0.84 -0.02 0.14 0.88PBD (max) 0.06 0.05 0.20 0.07 0.05 0.18BE F (y/n) 0.02 0.10 0.87 0.01 0.11 0.93 0.02 0.11 0.89BE M (y/n) -0.12 0.10 0.23 -0.13 0.11 0.23 -0.15 0.11 0.21
101
* Estimates are relative to 1.+Experienced breeders are those that fledged at least one offspring before (1 = experienced breeder, 0 = first-time breeder).
(C) Hatching successn98, 40F, 44M, 6YAIC 143 143.40 145.90 146.20 147.00 149.00Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 2.51 0.42 <0.001 2.60 0.42 <0.001 1.94 0.72 <0.001 2.54 0.43 <0.001 1.93 0.69 0.01 1.89 0.78 0.02Territory quality -0.06 0.25 0.79 -0.05 0.23 0.83 -0.08 0.24 0.75 0.07 0.25 0.80 0.04 0.26 0.88 0.04 0.26 0.89Group size 0.14 0.26 0.59 0.08 0.26 0.76 0.05 0.27 0.87 0.04 0.29 0.90 0.01 0.29 0.98 0.01 0.29 0.99PBD 0.19 0.71 0.78 -0.40 0.89 0.65 0.50 0.71 0.48 -0.03 0.34 0.92 -0.12 0.96 0.90PBD ^2 -0.46 0.67 0.49 -0.09 0.77 0.91 -0.31 0.66 0.64 0.08 0.78 0.92PBD (max) 0.81 0.39 0.04 0.84 0.40 0.03BE F (y/n) 0.79 0.65 0.23 0.75 0.64 0.24 0.78 0.72 0.28BE M (y/n) 0.21 0.59 0.72 0.19 0.61 0.75 0.22 0.65 0.74
(D) Fledging success n97, 48F, 47M, 6YAIC 204.60 205.3 205.10 206.60 207.10 209.10Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 1.20 0.30 <0.001 1.17 0.32 <0.001 0.93 0.51 0.07 1.19 0.30 <0.001 0.92 0.54 0.09 0.91 0.55 0.09Territory quality -0.04 0.15 0.82 -0.02 0.14 0.88 -0.02 0.15 0.92 -0.03 0.15 0.82 -0.02 0.15 0.92 -0.01 0.15 0.94Group size 0.01 0.17 0.96 0.12 0.17 0.50 0.03 0.18 0.88 0.01 0.17 0.96 -0.03 0.18 0.88 -0.03 0.18 0.87PBD 0.56 0.52 0.28 0.30 0.25 0.22 0.56 0.53 0.29 0.28 0.66 0.68 0.28 0.66 0.68PBD ^2 -0.17 0.57 0.76 -0.17 0.57 0.76 0.02 0.64 0.97 0.03 0.64 0.96PBD (max) -0.01 0.18 0.99 -0.02 0.18 0.93BE F (y/n) -0.13 0.40 0.75 -0.13 0.41 0.76 -0.12 0.42 0.77BE M (y/n) 0.50 0.41 0.22 0.51 0.43 0.24 0.51 0.44 0.24
102
Table 2. Effect of pair bond duration (PBD, in months) and individual breeding experience (BE, as factor+) on annual fledgling production (#
fledglings / year). Reproductive estimates from 2005-2015. B = # breeding pair-years, F = # females, M = #males, Y = # years. Random
factors (female and male identity and year) did not contribute to variation in any model. The top models (< 2 AIC) are shown, p < 0.05 in
bold
* Estimates are relative to 1.+Experienced breeders are those that fledged at least one offspring before (1 = experienced breeder,0= first-time breeder).
(A) Annual fledglingproduction B201, 84F, 95M, 10YAIC 635.2 638.10 639.80 639.90 641.50 643.20Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 0.63 0.09 <0.001 0.63 0.09 <0.001 0.63 0.09 <0.001 0.69 0.13 <0.001 0.70 0.13<0.001 0.70 0.13 <0.001Territory quality 0.1 0.05 0.05 0.09 0.05 0.07 0.09 0.05 0.09 0.09 0.05 0.08 0.09 0.05 0.08 0.09 0.05 0.09Group size 0.08 0.05 0.12 0.06 0.06 0.26 0.06 0.06 0.25 0.06 0.06 0.26 0.07 0.06 0.22 0.07 0.06 0.22PBD 0.10 0.14 0.47 0.06 0.16 0.73 0.07 0.06 0.25 0.16 0.16 0.33 0.11 0.18 0.54PBD ^2 -0.05 0.14 0.70 -0.04 0.14 0.77 -0.09 0.15 0.55 -0.08 0.15 0.62PBD (max) 0.05 0.08 0.56 0.04 0.08 0.61BE F (y/n) -0.07 0.13 0.61 -0.08 0.14 0.54 -0.09 0.14 0.55BE M (y/n) -0.02 0.14 0.86 -0.04 0.14 0.80 -0.03 0.14 0.84
103
Table 3. Longer pair bonds (PBD max, in months) tended to be associated with territories of
higher quality, although very long pair bonds had territories of lower quality (i.e. quadratic
effect). Each pair is included once, since pairs hold the same territory for several years. Males
and females were found in different pair combinations and included as random terms; these
did not contribute to variance of model. P < 0.05 in bold
Territory quality n = 89Fixed effects Estimate SE p
Intercept 9.66 0.5 <0.001Group size 1.19 0.50 0.02PBD (max) 2.90 1.61 0.07PBD (max) ^2 -3.48 1.61 0.03
104
Table 4. Longer previous pair bond duration (PBD, in months, census 1 July year x) was not
associated with greater likelihood (1/0) of surviving the following 12 months (to 1 July year x
+ 1) for males (M) or females (F). Individual breeding experience (BE, as factor+) in females
tended to be associated with a lesser likelihood of surviving the following 12 months,
possibly an effect of age. Results from a GLM with binomial distribution. Estimates are
relative to 1, n = # individuals. p < 0.05 in bold
Females MalesModels n= 67 n = 77Fixed effects Estimate SE p Estimate SE p
Intercept 2.28 0.58 <0.001 2.18 0.57 <0.001Fledgling production 0.07 0.25 0.77 0.52 0.32 0.10Territory quality -0.04 0.25 0.88 0.27 0.30 0.38Group size 0.46 0.32 0.15 -0.49 0.32 0.13PBD 0.52 0.81 0.52 -2.02 1.10 0.07PBD ^2 -1.02 0.78 0.19 1.95 1.11 0.08BE F (y/n)*+ -1.31 0.70 0.06BE M (y/n)*+ -0.03 0.75 0.97
* Estimates are relative to 1.+Experienced breeders are those that fledged at least one offspring before (1 = experienced breeder,0= first-time breeder).
105
Figure 1. The likelihood of nest success (success = yes, predation = no) increases with longer
pair bonds, however this effect decreases with pair bonds that last very long. This effect
is independent of individual breeding experience. Shown are predicted values of a
generalized linear mixed model. Other fixed terms were territory quality and group size.
Model shown includes pair bond duration as a quadratic term, there were no effects of
maximum pair bond duration as a covariate (to control for selective disappearance).
Nes
t suc
cess
Pair bond duration (months)
106
Figure 2. Territories of higher quality are held by pairs that stay together longer (positive
linear effect p = 0.07) however this effect decreases with pair bonds that last very long
(negative quadratic effect, p = 0.03). Territory quality is an index of Pandanus vegetation
cover ranging from 1-20. Shown are raw values, other covariate in generalized linear
mixed model was group size. Red line shows best fit to raw data and dotted lines show
intervals of confidence, green line shows the best linear fit. Males and females were
found in different combinations and included as random terms that did not contribute to
the variance of the model.
Pair bond duration (months)
Terr
itory
qua
lity
107
Supplementary material
Table S1. Effect of pair bond duration (PBD) and individual breeding experience (BE as continuousvariable) on productivity. (A) Probability of nest success (predation = 0, success = 1). (B) Clutch size,(C) hatching success (proportion eggs hatching) and (D) fledging success (proportion chicks fledging)(E) annual fledgling production. Reproductive estimates from 2005-2010 for A-D, until 2015 for E. n= # nests, F = # females, M = #males, Y = # years, B = # breeding pair-years. Random factors (femaleand male identity and year) did not contribute to variation in any model. Models arranged by AICvalues; all models shown.
(A) Nest success n78, 31F, 37M, 6YAIC 119.4 121.4 122.6 123.6 124.5Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 0.08 0.24 0.72 0.09 0.24 0.72 0.08 0.23 0.73 0.09 0.24 0.72 0.08 0.23 0.73Territory quality 0.02 0.24 0.91 0.03 0.25 0.90 0.06 0.27 0.84 0.03 0.25 0.90 0.06 0.28 0.84Group size 0.43 0.32 0.19 0.43 0.33 0.19 0.37 0.31 0.23 0.43 0.33 0.19 0.38 0.31 0.23PBD 1.21 0.78 0.12 1.20 0.80 0.13 0.48 0.58 0.41 1.21 0.79 0.13 -0.48 0.58 0.41PBD ^2 -1.34 0.78 0.08 -1.35 0.78 0.08 -1.35 0.78 0.08PBD (max) 0.02 0.32 0.93 0.03 0.32 0.93Breeding experience F -0.03 0.38 0.95 -0.03 0.38 0.95Breeding experience M 0.44 0.46 0.34 0.44 0.47 0.34
(B) Clutch size n112, 36F, 39M, 6YAIC 415.7 417 417.7 419.6 421.1Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 1.25 0.05 <0.001 1.25 0.05 <0.001 1.25 0.05 <0.001 1.25 0.05 <0.001 1.30 0.05 <0.001Territory quality 0.02 0.05 0.66 0.01 0.05 0.78 0.01 0.06 0.81 0.02 0.06 0.79 0.01 0.06 0.86Group size 0.09 0.06 0.16 0.09 0.06 0.15 0.08 0.06 0.17 0.08 0.06 0.20 0.08 0.06 0.19PBD -0.07 0.15 0.64 -0.10 0.15 0.52 -0.03 0.11 0.75 -0.07 0.17 0.67 -0.11 0.18 0.55PBD ^2 0.04 0.14 0.78 0.04 0.14 0.78 0.04 0.14 0.77 0.04 0.14 0.78PBD (max) 0.05 0.07 0.44 0.05 0.07 0.46Breeding experience F 0.02 0.08 0.82 0.02 0.08 0.82 0.01 0.08 0.88Breeding experience M -0.01 0.09 0.90 -0.02 0.08 0.82 -0.01 0.09 0.89
(C) Hatching success n59, 26F, 31M, 6YAIC 93.30 93.90 94.80 95.0 95.30Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 2.72 0.56 <0.001 2.83 0.59 <0.001 2.73 0.57 <0.001 2.85 0.59 <0.001 2.70 0.55 <0.001Territory quality -0.20 0.39 0.75 -0.13 0.39 0.74 0.11 0.37 0.76 -0.14 0.37 0.7 0.06 0.33 0.85Group size 0.14 0.40 0.73 0.25 0.44 0.57 0.40 0.45 0.38 0.27 0.45 0.54 0.40 0.44 0.38PBD -0.50 0.78 0.52 0.41 1.10 0.71 0.80 0.93 0.39 0.17 1.11 0.88 0.55 0.97 0.30PBD ^2 -1.06 0.89 0.23 -0.88 0.93 0.35 -1.18 0.91 0.20 -0.95 0.92 0.30PBD (max) 0.53 0.57 0.34 0.65 0.54 0.23Breeding experience F 1.29 0.87 0.14 1.45 0.96 0.13 1.43 0.97 0.14Breeding experience M -0.43 0.41 0.30 -0.49 0.43 0.25 -0.37 0.42 0.38
(D) Fledging success n56, 29F, 33M, 6YAIC 128.7 129.1 130.7 132.7 204.6Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 0.98 0.25 <0.001 0.95 0.19 <0.001 0.99 0.26 <0.001 0.99 0.27 <0.001 1.20 0.30 <0.001Territory quality 0.17 0.21 0.41 0.13 0.19 0.50 0.17 0.21 0.41 0.17 0.21 0.41 -0.03 0.15 0.82Group size 0.46 0.29 0.11 0.43 0.27 0.11 0.45 0.29 0.12 0.45 0.30 0.13 0.01 0.18 0.96PBD 0.55 0.44 0.22 0.29 0.68 0.67 0.44 0.80 0.58 0.44 0.83 0.59 0.56 0.52 0.28PBD ^2 0.08 0.82 0.93 0.13 0.87 0.88 0.13 0.89 0.88 -0.17 0.57 0.76PBD (max) -0.02 0.23 0.92 0.01 0.26 0.99Breeding experience F -0.24 0.44 0.59 -0.25 0.44 0.57 -0.25 0.48 0.60Breeding experience M 0.12 0.36 0.59 0.12 0.36 0.75 -0.11 0.36 0.75
(E) Annual fledglingproduction B153, 68F,77M, 10YAIC 495.3 496.6 497.2 502.7 504.1Fixed effects Estimate SE p Estimate SE p Estimate SE p Estimate SE p Estimate SE pIntercept 0.63 0.09 <0.001 0.63 0.09 <0.001 0.63 0.10 <0.001 0.62 0.09 <0.001 0.62 0.09 <0.001Territory quality 0.08 0.06 0.22 0.08 0.06 0.20 0.08 0.06 0.19 0.07 0.06 0.22 0.07 0.06 0.22Group size 0.10 0.07 0.14 0.11 0.07 0.13 0.10 0.07 0.15 0.09 0.06 0.13 0.09 0.06 0.13PBD -0.10 0.16 0.51 0.02 0.22 0.91 -0.12 0.25 0.65 0.16 0.16 0.36 0.08 0.20 0.68PBD ^2 -0.15 0.17 0.40 -0.11 0.18 0.53 -0.16 0.17 0.36 -0.13 0.18 0.44PBD (max) 0.11 0.10 0.24 0.07 0.09 0.46Breeding experience F -0.02 0.11 0.86 -0.02 0.11 0.83 -0.04 0.11 0.74Breeding experience M 0.12 0.12 0.32 0.13 0.12 0.28 0.17 0.13 0.19
108
Table S2. Longer pair bond duration (PBD) was not associated with greater annual likelihood
of survival (1/0) for males (M) or females (F). Survival of females was lower age. The last
two years of records for each breeder with known pair bond duration were included. Results
from a GLM with binomial distribution. Estimates are relative to 1, n = # individuals. P < 0.05
in bold
Females MalesModels n67 n77Fixed effects Estimate SE p Estimate SE p
Intercept 2.7 0.81 <0.001 2.2 1.25 0.08Fledglingproduction 0.09 0.27 0.73 0.50 0.33 0.13Territory quality 0.06 0.28 0.84 0.18 0.32 0.56Group size 0.34 0.35 0.33 -0.59 0.35 0.09PBD 0.27 0.81 0.74 -1.72 1.01 0.09PBD ^2 -0.52 0.75 0.50 1.72 1.10 0.12Age F -0.03 0.01 0.08 -0.01 0.02 0.97Age M
109
Figure S1. Number of partners individuals had throughout their lives (n = 317 pairs). All
pairs in our population included which are, pairs with known and unknown pair bonds (pairs
found at original study population in 2005) and incestuous pairs (n = 29).
91
49
206 2 0 4
124
3319 12
2 1 10
20
40
60
80
100
120
140
1 2 3 4 5 6 7
Males Females
110
Declaration for Thesis Chapter 4
Declaration by candidate
In the case of Chapter 4, the nature and extent of my contribution to the work was the
following:
Nature of contribution Extent of contribution (%)
The key ideas, data analysis, lab work and writing of the
manuscript were my responsibility.
90
The following co-authors contributed to the work. If co-authors are students at Monash
University, the extent of their contribution in percentage terms must be stated:
The undersigned hereby certify that the above declaration correctly reflects the nature and
extent of the candidate’s and co-authors’ contributions to this work.
Candidate’s Signature Date: 15-10-2016
Main Supervisor’s Signature Date: 15-10-2016
Name Nature of contribution
Anne Peters Assistance with writing manuscripts, design of hypotheses, data
collection, interpretation of results, feedback on thesis
Michelle L. Hall Assistance with writing, data collection, interpretation of results.
Sjouke A. Kingma Assistance with writing, data collection, interpretation of results.
Paul Sunnucks Assistance with writing, interpretation of results.
111
Chapter 4
Incest avoidance, extra-pair paternity and territory quality
drive divorce in a year-round territorial bird
Nataly Hidalgo Aranzamendi1, Michelle L. Hall2,3, Sjouke A. Kingma3,4, Paul
Sunnucks1 and Anne Peters1,3,5
1School of Biological Sciences, Monash University, Clayton, Victoria, Australia
2School of BioSciences, University of Melbourne, Melbourne, Victoria, Australia
3 Max Planck Institute for Ornithology, Vogelwarte Radolfzell, Radolfzell, Germany
4 Behavioural and Physiological Ecology, Groningen Institute for Evolutionary Life Sciences,
University of Groningen, Groningen, The Netherlands
5 Australian Wildlife Conservancy, Australia
Behavioral Ecology doi. 10.1093/beheco/arw101
Abstract
Divorce can be an important behavioral strategy to improve fitness. This is particularly
relevant for species that are territorial year-round with continuous partnerships, where
individuals face constraints on partner choice due to limited vacancies and dispersal
opportunities. We tested several hypotheses for divorce in such a species, the cooperatively
breeding bird Malurus coronatus. Based on nine years of detailed information on dispersal
and survival of 317 breeding pairs, we tested if divorce is driven by inbreeding avoidance, by
a better partner or territory, or by social variables (number of subordinates and fidelity of
partners). We found that divorce is important to escape incest: incestuous pairs were
substantially more likely to divorce (64%) than non-incestuous pairs (14%). However,
incestuous pair bonds lasted up to a year, highlighting constraints on breeder dispersal. Non-
incestuous pairs also divorced, but here the only predictor for divorce was the presence of
112
extra-pair offspring in a previous brood. Although reproductive failure did not trigger
divorce, and reproductive success did not improve in the year following divorce, females that
dispersed after divorce obtained higher quality territories, unlike females that dispersed after
their partner died. Thus divorce may be a strategy to improve long-term benefits associated
with better territories. Some divorces appeared to be forced evictions by older females,
although direct evidence for this is limited. Taken together, our findings demonstrate the
complexity of factors that affect the occurrence of divorce when partner choice is
constrained.
Keywords: breeder dispersal, inbreeding, infidelity, long-term pair bond, monogamy,
cooperative breeding
Introduction
In pair-breeding animals, optimal choice of a breeding partner has major fitness benefits
(Andersson 1994, Black et al. 1996). Additionally, there is ample evidence that mate
familiarity results in higher fitness over time, especially in species that establish long-term
monogamous pair-bonds and have bi-parental care (Black et al. 1996, van de Pol et al. 2006).
Nonetheless, divorce, the dissolution of a pair bond in which both individuals remain alive
and one or both have re-paired, occurs in 92% of socially monogamous birds (Choudhury
1995, Black 1996, Ens et al.1996, Culina et al. 2014). It is therefore expected that changing
social partners and re-mating should be an adaptive decision, with benefits greater than costs
(Culina et al. 2014). In species where partners separate between breeding seasons, the
benefits of fidelity and the costs of finding a new partner are relatively easily surpassed by
the costs of waiting for the former partner who might not return (Ludwig and Becker 2006).
However, in species that form year-round pair bonds, there is no uncertainty over the fate of
the partner. In such species, re-pairing after divorce requires eviction of another breeder or
113
opportunistic dispersal into a vacancy created by mortality or dispersal of territory owners
(Ens et al. 1993, Dubois et al.1998). Thus, in such systems, divorce is expected to represent a
balance between abandoning a breeding position (i.e. cost), and the relative benefits that are
associated with doing so.
Several adaptive hypotheses for causes of divorce have been proposed (reviewed by
Choudhury 1995). For species with year-round bonds they can be grouped in three main
categories: (a) the ‘Better-Option’ Hypothesis (Davies 1989, Ens et al. 1993, Desrochers and
Magrath 1996) predicts that individuals with low prior breeding success, a low quality
partner, or in a low quality territory might move to opportunistically take a better option. This
hypothesis predicts that at least one member of the divorced pair will benefit by increasing
their reproductive success with a new partner. This could be achieved by obtaining a more
compatible or higher quality partner or a better breeding territory. (b) The Inbreeding
Avoidance Hypothesis (Koenig and Haydock 1998, Hatchwell et al. 2000, Cockburn et al.
2003) predicts that divorce is more likely when partners are closely related, which is
particularly relevant for species with kin-structured populations, in which such pair bonds are
relatively common due to inheritance of a breeding position. If so, most likely both, but at
least one, members of the pair will re-pair with an unrelated individual after divorce. (c) The
Forced-Divorce Hypothesis (Taborsky and Taborsky 1999) predicts that divorce is not
initiated by either partner but caused by an intruder that displaces a member of the same sex,
and fitness benefits are accrued by the intruder and no improvement will be evident for
former pair members after divorce. These hypotheses are not exclusive and multiple
mechanisms may occur within one species. Therefore, hypotheses should be considered
simultaneously when assessing the underlying causes of divorce (Choudhury 1995).
A recent meta-analysis (Culina et al. 2014) confirmed that divorce is an adaptive
strategy to increase reproductive success, rather than an unselected, non-adaptive event or a
114
by-product of another strategy (Choudhury 1995). This same study also showed that the
social and ecological factors driving divorce remain elusive (Culina et al. 2014). A
contributing aspect may be that factors driving divorce are often studied independently of
factors driving dispersal, yet the two processes are usually inextricably linked. In year-round
territorial species, divorce is not just about changing partners, but requires one individual to
disperse - with all the associated costs and benefits. Limited understanding of dispersal
patterns can lead to a failure to distinguish emigration and mortality, with non-breeding
individuals that dispersed to outside the study site assumed dead (Culina et al. 2015). In such
cases, the observed divorces are a subset of true divorces. A related limitation frequently
encountered is the inability to differentiate between two re-pairing categories: divorced and
widowed (Culina et al. 2014). For example, 11 of 16 studies in Culina et al. (2014, Table S1)
considered both categories as one: birds that changed partners. However, fitness benefits
should be most evident when contrasting re-paired divorced with re-paired widowed
individuals (Ens et al. 1993, Culina et al. 2014), to distinguish between the actual outcomes
of divorce versus partner change. Moreover, other life-history strategies like mating outside
the pair bond (i.e. extra-pair mating) might affect selection pressures to divorce. Extra-pair
mating and divorce can be alternative or complementary strategies for mate choice, both with
a common target, which is to improve reproductive success (Birkhead and Møller 1992,
Cézilly and Nager 1995, Ramsay et al. 2000). Thus, in order to understand the adaptiveness
of divorce, studies should consider the incidence of alternative mating strategies, in a species
with reliable longitudinal information on survival, post-divorce dispersal and new pairing
status. This will allow to differentiate between divorced, widowed and faithful pairs (as in
Ens et al. 1993, Dhondt and Adriaensen. 1994, Orell et al. 1994, Jeschke et al. 2007, Culina
et al. 2014). Our study of divorce in a species with continuous partnerships and year-round
territoriality fulfils these criteria.
115
Here we investigate factors underlying divorce and its consequences in the purple-
crowned fairy-wren (Malurus coronatus), a year-round territorial cooperatively breeding
passerine. We test which predictions of the major hypotheses for divorce (Table 1) are
supported in this species that forms long-term pair bonds, can breed year-round and lives in
kin-based social groups in saturated habitats with constraints on pairing opportunities
(Kingma et al. 2009). We show that occurrence of divorce cannot be explained by a single
hypothesis, and several predictions were supported.
Methods
Study species
A population of M. coronatus was monitored at Mornington Wildlife Sanctuary, northwest
Australia (17°31’S, 126°6’E) from July 2005 until June 2014. The core study area is located
along ~15 km of river length between Annie Creek and the Adcock River. Cooperative
groups - a dominant breeding pair accompanied by a variable number of male and female
subordinates - defend year-round territories linearly aligned along creeks and rivers in
vegetation dominated by Pandanus aquaticus, an evergreen palm-like medium-sized tree
(Rowell and Russell 1997, Kingma et al. 2009). Only the dominant pair performs duets and
reproduces and male and female subordinates can help raise offspring, improving
productivity and survival of dominants (Hall and Peters 2008, Kingma et al. 2010). Breeding
can occur in every month of the year, with a peak during the wet season (Dec-Mar) and often,
but not always, a peak during the dry season (Sep-Oct) (Rowley and Russell 1997, Peters et
al. 2013). Although the population-wide incidence of extra-pair paternity (EPP) is only 5%,
incestuous pairs have a much higher rate of EPP (46%) compared with 3.3% among non-
incestuous pairs (Kingma et al. 2013). Some subordinates might be unrelated to one or both
members of the dominant pair (Kingma et al. 2011b), but the majority (> 60%) are retained
116
offspring, and incestuous pairs result from subordinates inheriting a breeding vacancy after
the disappearance of the same-sex breeder (Kingma et al. 2011b). Overall, natal dispersal is
female-biased with most subordinate males remaining in their natal territory or moving
nearby, while females generally disperse farther; however, some subordinate females also
show natal philopatry (Kingma et al. 2010, 2013).
Field data collection
Territory boundaries were recorded from repeated observations of the birds’ movements and
the location of agonistic interactions between groups. Most boundaries remained stable
through the years but occasional shifts were recorded. Pandanus cover was quantified as an
index of territory quality. Birds do not occupy stretches of river vegetation without Pandanus
and they depend strongly on this vegetation (51% day-time spent in Pandanus, and 95% of
nests built in Pandanus) (Kingma et al. 2011a). Territories with greater Pandanus cover have
more subordinates (Kingma et al. 2011a), indicating that these territories are more productive
or more attractive for subordinates. Because subordinates increase breeder survival and
fledgling recruitment (Kingma et al. 2010), territories with greater Pandanus cover thus
provide benefits to breeders. Also, higher Pandanus cover reduces the likelihood of nest
predation, which is the most important source of reproductive failure (Hidalgo Aranzamendi
et al. unpubl. data). Thus, we assume that a territory with greater Pandanus cover reflect the
quality of a breeding territory. Pandanus cover was quantified along the territory by one
observer assigning a score between 1-10, every 50 m (10 each river side, maximum total 20).
An average of all points was calculated for territories longer than 50 m.
Since 2005, all birds in the core study area were individually color-banded and group
composition, dispersal and survival documented. Between 2005 and 2010, territories were
monitored weekly year-round and all breeding attempts followed. From 2010 onwards, the
117
population was monitored in two censuses per year: in May-June and in November, following
the reproductive peaks. During this period, all new unbanded birds were banded and a blood
sample was taken for parentage analysis. All banded birds that survived were resighted and
social status (subordinate or dominant) recorded.
From 2007 onwards intensive yearly censuses along the tributaries that join the study
site were conducted, to find birds that had dispersed outside the core area (emigrants). These
censuses covered almost all suitable habitat within 20 km of the core area, and some up to 60
km away, covering a total of 95 km of river length (including Adcock, Hann, Fitzroy rivers
and Throssell, Roy and Spider creeks). To locate birds during these censuses, a 90-second
song playback was used at intervals of 100 m or less to attract resident birds. The accuracy of
this technique is high, since 90% of dominant birds respond to on-territory playback (Hall
and Peters 2008). During these censuses, eight divorced breeders were found (plus 22 other
banded birds from the core area). Additionally, no bird from our population was resighted
during censuses conducted at a landscape scale in the surrounding potential habitat, and
dispersal outside the river vegetation has never been recorded for this species (Skroblin and
Legge 2010, Skroblin et al. 2014).
Between 2005 and 2010, no dominants were resighted after being declared dead on
the basis of failure to sight them in intensive surveys. After 2010, 3.2% of breeders were
initially assumed dead in one census but then found during a subsequent census in a different
territory within the core area (n = 10 birds). Given that no birds were resighted outside the
core area after being declared dead, it can be assumed that very few pair bonds classified as
ended by death actually were divorces.
Throughout the study, 317 different breeding pairs were recorded, totaling 507 pair-
years (179 dominant males, 192 dominant females). Average group size (breeders plus
subordinates) was stable around a mean of 3.6. Annual survival of dominants was on average
118
81% and average sex-ratio of adults was 1.2 males per female (Kingma et al. 2009). On
average, 7.3 (range 6-10) new immigrant females entered the study site per year. Most (70%)
of these settled as dominant breeders, totaling 6.9% of immigrant breeders per year of the
study period (no. immigrant birds*100/no. breeding adults). The proportion of immigrants
did not differ between sampling periods (mean before 2010 = 7.8, mean after 2010 = 8.0
immigrants/year). The number of fledglings detected per year did not differ between the two
study periods (mean before 2010 = 0.55 fle/pair/year, mean after 2010 = 0.57 fle/pair/year, t-
test χ 2 = -0.18, df = 1, p = 0.43)
Data analysis
Pair bond duration and frequency of divorce- Divorce was defined as the failure to maintain
a pair bond when both partners were still alive and observed with new social partners and/or
in new locations. Annual divorce rate was calculated for July – June by dividing the total
number of divorces by the total number of pairs present in that year.
Until 2010, pair-bond duration (PBD) was calculated as the time (days) between the first and
last date a pair was seen together. After 2010, the start date of a partnership was estimated as
the midpoint date between two censuses when birds were first seen paired and the previous
census; the end date of a partnership was estimated as the midpoint date between the last
census birds were seen paired and the subsequent census. To assess how accurate our
estimates after 2010 were, we recalculated PBD for pairs that started and ended between
2005-2010 using the same methods as 2010-2015, by simulating two censuses per year and
assigning midpoint dates as the start and end of the relationships. Following this method, we
would have missed 15 relationships that lasted less than six months (5 divorces and 10
deaths), out of the 195 relationships in total (7.7%). The simulated estimates of PBD closely
resemble the PBD based on weekly censuses (r = 0.98), with a mean difference of 43 days
119
(median = 38). Considering that the mean PBD for all relationships before 2010 was 370
days (median = 201) we believe that the less intensive sampling after 2010 does not alter our
conclusions markedly.
Paternity analysis- DNA was extracted following Barrett et al. (2012). Paternity was assigned
using Cervus 3.0.3 (Marshall 1998, Kalinowski et al. 2007). Throughout the study, 1111
individuals were genetically screened by two commercial suppliers using 6 or 9
microsatellites (as in Kingma et al. 2009, 2013). Until 2010, we sampled nestlings and these
could be unambiguously assigned genetically to the local breeding female. The sires of EP
offspring were identified by repeating the analysis including as possible fathers all dominant
and subordinate males present at the time of capture. Less than 5.7% offspring were sired by
an extra-pair (EP) father (Kingma et al. 2009, 2013). After 2010, nestlings were no longer
sampled and instead we assigned paternity to unbanded birds found at each census (these
could be fledglings, subordinates or immigrants). First, all unbanded birds were tested as
offspring of the local breeding pair (i.e. the dominant pair of the territory in which they were
captured). If the local breeding pair did not match as genetic parents, we tested whether the
unbanded birds were subordinates that had already moved from their natal territory; we did
this by testing as potential parents all breeding pairs present at the time of capture or last seen
in the previous census (i.e. each pair in the population was tested as putative parents to
previously unassigned birds). We also tested EP fathers as indicated above. Unbanded birds
were considered immigrants if they could not be assigned to the local breeding pair or to
other breeding pairs. The two periods with different methods of assigning parentage yielded
comparable results: 2011-14, the percentage of EPP remained similar (6.1%) compared to
during 2005-2010 (5.7%; Kingma et al. 2013).
Incest and relatedness- Information about the occurrence of incest was extracted from a
genetic pedigree containing all known parent-offspring relationships. Twenty-nine socially-
120
bonded pairs were thus identified as first-degree relatives (mother-son, father-daughter, and
brother-sister), and categorized as “incestuous”. One brother-sister pair was excluded from all
analyses because they were unfamiliar with one another (hatched in different years and never
concurrently resident on their natal territory). All other pairs were categorized as “non-
incestuous”; according to the pedigree, five of those pairs consisted of second- and third-
degree relatives. In addition, pairwise genetic relatedness of pair members (LR estimator;
Lynch and Ritland 1999) was calculated using 6 microsatellites (as in Kingma et al. 2013)
with GenAlEx 6.5 (Peakall and Smouse 2012).
Statistical Analyses
Design of analyses
The hypotheses explaining adaptive divorce make several different predictions (see Table 1;
adapted from Choudhury 1995). The Inbreeding Avoidance hypothesis predicts that divorce
is more likely when partners are related (e.g. incestuous pair, first-order relatives). We define
inbreeding avoidance here as avoiding mating with first-order relatives (i.e. incest avoidance,
Koenig and Haydock 1998). This hypothesis predicts that one or most likely both partners
should show reduced relatedness to their new partner after divorce. The Better Option
hypothesis predicts that individuals with low prior breeding success, a low quality partner, or
in a low quality territory might move to opportunistically take a better option. The benefits of
divorce should be evident in the outcome after divorce for one of the former partners, if they
achieve greater reproductive success, obtain a better territory or a better partner than
previously. Additionally, divorce can also be associated with infidelity (extra-pair mating). In
this case, it is predicted that either females will divorce and subsequently pair with a former
EP mate or males will evict unfaithful females. The predictions of the Forced Divorce
hypothesis are clear: the only partner characteristics that might be identified as drivers of
121
divorce would be lower competitive ability than the replacement individual (age, size). After
divorce, neither ex-partner is expected to benefit, only the intruder. In this case, previous
reproductive success does not affect the probability of divorce, and there is no change in
breeding success after divorce.
To test these predictions, (1) we first compared the pattern of pair bond persistence
for incestuous and non-incestuous pairs. Divorce affected those two groups differently to
such an extent that they could not be analyzed in a single statistical model (all incestuous
pairs divorced within a year; see results); so, for all subsequent analyses, incestuous and non-
incestuous pairs were analyzed separately. (2)We analyzed the possible drivers of divorce
and pair bond duration, and (3) we analyzed the outcomes of divorces to test if divorce led to
improvements of breeding success or reduced relatedness to partners in either sex. Finally,
we explored if any of the patterns found were consistent with forced divorces.
Samples sizes vary for each analysis due to missing parameters, see Tables for sample
sizes and text for details.
1. Pair bond persistence: the importance of incest avoidance
To investigate to what extent inbreeding avoidance is a driver of divorce, we tested whether
the probability of divorce depended on the type of relationship (incestuous or not;
independent variable) using a survival analysis. This allows inclusion of pairs in which an
individual died and persisting pairs at the end of the study (as right-censored cases). This
analysis expresses the probability that relationships lasting any given length will be
terminated by divorce (Cox 1972, Cockburn et al. 2003). First, the persistence over time of
all incestuous and non-incestuous pairs in each of the three classes (divorced, widowed and
ongoing) was plotted using Kaplan-Meier graphs with the survdiff function of the survival
package (Diez 2012). Second, a Cox’s proportional hazard analysis was performed
comparing pair bond persistence in both groups (using coxph in the same package).
122
2a. Drivers of divorce: What determines the likelihood of pair bond dissolution?
To investigate which factors are drivers of the likelihood of divorce in non-incestuous pairs,
we scored for each year of study (July-June) whether pairs present on 1st July of each year
were still together or ended in divorce (binomial response variable). This dataset (n = 126
pairs) excluded relationships ending due to death of one of the pair-members, because these
could have divorced before death, if both had survived. For this, a GLMM was constructed
including male ID and female ID as random factors, to account for multiple entries for males
and females that stay together longer than a year or that were in more than one pair over the
study. Predictor variables were: genetic relatedness (LR estimator), reproductive success,
territory quality and group size; we also controlled for female origin (local/immigrant; all
males are residents). Group size (on 1st July of each year) was included, because breeders
derive reproductive and survival benefits from living in larger groups (Kingma et al. 2011b).
Similarly, a territory of high or low quality could be an asset that influences divorce (see
Field Data Collection for details). Reproductive success (number of fledglings, range 0-8)
was the number of offspring that survived at least six weeks during the calendar year (July-
June). Annual variation in fledgling production is larger (range 16-37 fle/year) than the
variation between the two study periods (see Field Data Collection), therefore year was
included as a random factor.
We built a second GLMM with the predictor and random variables of the first model
adding another predictor variable: whether any previous broods of the pair contained EP
offspring (as a proxy of infidelity in the pair). For this analysis, only pairs that reproduced
and for which offspring had been genotyped could provide this information, so sample size
was much reduced (n = 53 pairs).
Since the number of fledglings produced has a large degree of stochasticity (e.g. nests
often fail due to predation and flooding, unpubl. data), this reduces the likelihood of detecting
123
breeding failure as a driver of divorce (Culina et al. 2014). For this reason, and because birds
can use hatching success as an indicator of within-pair compatibility, another model was built
using hatching success (proportion of eggs that hatched from each clutch for all nests found
in a calendar year) instead of number of fledglings. Hatching success data were available
only for 2005-2010, when all nesting attempts were followed in detail. Presence of EP
offspring was not included, due to insufficient sample size (we did not know paternity for a
large number of nestlings that were depredated or nests flooded before sampling).
Although clutch size and timing of breeding may be important determinants of fitness
in general (Culina et al. 2014), we did not test whether these factors affected the likelihood of
divorce in M. coronatus. Clutch size is extremely consistent (2.93±0.05 SE eggs, unpubl.
data) and birds breed year-round, thus the definition of timing of the onset of breeding would
be arbitrary and but more importantly, its meaning would be unclear in terms of fitness, as we
have no indication that there are distinct seasonal differences in success.
2b. Pair bond duration in divorced pairs
To test if any factor predicted pair bond duration (number of days, log-transformed,
dependent variable) for pairs that eventually divorced, we built linear models with genetic
relatedness, territory quality and group size as independent variables. We constructed
separate models for incestuous and non-incestuous pairs. Genetic relatedness was not
included in the model of incestuous pairs. Group size at the start of the pair bond was used to
avoid problems distinguishing cause and effect (pairs that were together longer had more
time to breed and produce helpers).
3. Outcomes of divorce
Individuals may divorce in order to obtain a less related partner, or a partner of higher
quality, and/or, if an individual disperses, better quality of the new territory. Those changes
can only be detected by comparing partners and territories (former vs. new) and by
124
comparing individuals that have experienced partner change: divorcees vs. widows
(individuals whose partner died). Three approaches were used to test whether divorce
resulted in improvements in partner or territory quality: (i) within-bird comparison between
the new partner and the ex-partner (paired comparison) (ii) comparison post-divorce between
ex-partners and their new partners (paired comparison) and (iii) comparison of changes in
divorcees compared to changes for widows.
Predictor variables were: genetic relatedness, reproductive estimates, age and size of partners,
territory quality and group size. For analyses of reproductive estimates, we included the
offspring produced during the last year with the ex-partner versus the offspring of the first
year with the new partner. Pairs that were together less than 6 months or associated only
through the dry season were excluded, since the wet season is the time when most pairs
reproduce. Age and size were used as measures of individual quality or competitive ability.
Age has been reported as an important factor in divorce for other species (e.g. older partners
may be preferred after divorce or older birds may be less likely to divorce; McNamara and
Forslund 1996, Green et al. 2004, Cockburn et al. 2008, García-Navas and Sanz 2011).
Similarly, size is often regarded as an important signal of quality in small songbirds (Cardoso
2011); larger purple-crowned fairy-wrens males are able to produce lower-pitched low-
frequency songs, suggesting that size is important in sexual competition (Hall et al. 2013).
Tarsus length (±0.1 mm) was used as measure of body size. Finally, changes in territory
quality and group size are relevant only for individuals that dispersed (i.e. changed
territories).
(i) Within-bird comparisons were performed with the difference between the
characteristics of the new partner minus the old partner as a dependent variable in mixed
models, with bird ID as random factor. There is a significant difference between partners if
the intercept in these models differs from zero (Varian-Ramos and Webster 2012). We tested
125
whether new partners were less related, larger than previous partners or were older (including
only birds with known hatch date, age (in months) at the time of divorce vs age of new
partner at first date seen together), and if new pairs had higher reproductive success. To test if
post-divorce dispersal resulted in the acquisition of a higher quality territory (Pandanus
cover), or a larger group size, all female dispersers were grouped (incestuous and non-
incestuous) and the differences between the old and new (post-divorce) territory were
compared, adding bird ID as random factor. For males dispersing post-divorce, a paired t-test
(due to low sample size n = 7) was used to compare territory quality and group size between
the old and new territory. (ii) For post-divorce comparisons between ex-partners and their
new partners (divorced female and new partner vs divorced male and new partner), we
compared the differences in reproductive output between both new pairs using a GLMM
with: (number of fledglings produced by divorced female +new partner) – (number of
fledglings produced by divorced male+ new partner) as dependent variable and both male and
female ID as random terms. No other characteristic was compared between ex-partners and
their new relationships, since changes will be determined only by who dispersed or stayed
(analyzed in previous section). (iii) We compared changes in divorcees with changes for
widows, as a control group, to test costs of partner change. Genetic relatedness and
reproductive success of new pairs were compared between divorcees and widows. Similarly,
characteristics of new partners (age and size) were compared between divorcees and widows.
We used these variables as dependent terms in GLMMs with ML and Poisson distribution or
REML models for continuous variables, in which divorce and death were independent
variables and bird ID a random factor. Sample sizes for widowed incestuous pairs were small,
so we only compared non-incestuous pairs. To test if post-widowhood dispersal resulted in
changes in territory quality or group size, we compared the differences between the old and
new (post-widowhood) territory, this analysis was restricted to dispersing females (paired t-
126
test). Information of territory quality for males dispersing post-widowhood was incomplete (n
= 5 of 9 males dispersed to outside the area for which we had territory quality estimates), so
they were not analyzed.
All analyses were done in R 3.2.0 (R Core Team 2015). A mixed modelling approach
(GLMM) was used when data had repeated measures from the same individual(s) (as random
term(s)) using the lmerTest and lme4 package (Kuznetsova et al. 2014, Bates et al. 2015).
Generalized linear models (GLM) were run first without random terms and quasibinomial
distributions to estimate dispersion. All models indicated that data were not overdispersed (x
= 1.02), so GLMMs were appropriate (Quinn and Keogh 2002). R2 values (Nakagawa and
Schielzeth 2013, Johnson 2014) were calculated using the MuMIn package (Barton 2014).
For models with random terms, individual-level repeatability was calculated following
Nakagawa and Schielzeth (2010). Models had values of repeatability close to zero unless
reported. For model analyses without random effects, linear models (LM) and t-tests were
used. Residuals for models with continuous variables were checked for normality and
variables log-transformed if necessary; all independent variables were checked for
collinearity (all r < 0.56).
Results
Constraints on acquisition of first and subsequent partners
Almost 15% of males and 7% of females acquired their first partner by inheriting the
breeding position on their natal territory, and such inheritance generally resulted in the
formation of incestuous pairs (detailed information on routes to formation of first and
subsequent partners are provided in Table S1 and S2). In total, 28 incestuous pairs formed
during our study (first and subsequent partners): 15 males paired with their mother, 6 females
127
with their father, and 7 pairings were between full siblings. However, most individuals
acquired their first partner by natal dispersal: 60% females and 58% males.
Subsequent partners were obtained after a partnership ended in divorce (19% of 317
pairs recorded) or due to the death of one pair member (53% of 317 pairs); the remaining
partnerships ended when both partners died, 6%, or were ongoing, 22%. Eleven males and 8
females divorced more than once. The average annual rate of divorce was 8%, fluctuating
between 5% and 13%. Divorce occurred in any month of the year, although never when pairs
had dependent young.
Individuals that divorced and dispersed obtained an unrelated partner (except one
unusual case of pairing between unfamiliar relatives; Table S2). After divorce, in 88% of 50
cases, it was the female that dispersed much more frequently than males (binomial test p <
0.001). In contrast to divorcees, the majority of widowed individuals did not disperse (83%)
and this was similar for females (61 of 77) and males (78 of 87, binomial test p = 0.32). The
risk of pairing incestuously for birds that remained as breeder in the territory following
divorce or widowhood was higher for females (25% of 67) compared to males (8% of 119; χ2
= 9.88, df = 1, p = 0.001).
Divorce in incestuous pairs
Divorce occurred more often in pairs formed by first-order relatives (incestuous: 64% of 28)
compared with non-incestuous pairs (14% of 285) (Cox PH χ 2 = 18.50, df = 1, p < 0.001,
Fig. 1). All incestuous pairs divorced (n = 18, median = 107, range 7-364 days) or ended by
partner death (n = 10, median = 60, range 29-188 days) within a year. The duration of
incestuous pair bonds was not explained by territory quality (R2 = 0.21, F2,16 = 2.09, z =
0.05±0.06, p = 0.43), and although incestuous pairs in larger groups tended to stay together
128
longer than pairs in smaller groups, the effect size was small (z = 0.34±0.17, t = 1.97, p =
0.06).
Divorce in non-incestuous pairs
One driver of the likelihood of divorce was identified: non-incestuous pairs that previously
had EP offspring had a higher likelihood of divorce (z = 2.56±1.11, p = 0.02, Table 2). EP
offspring occurred in 21% of divorcing pairs (3 of 14 reproducing pairs, Fig. 2) compared to
8% in pairs that did not divorce (3 of 39, Fig. 2). Neither genetic relatedness between partners
nor previous reproductive success predicted whether pairs divorced (Table 2). Pair bond
duration for non-incestuous pairs that divorced was extremely variable, and sometimes very
long (n = 40, median = 224 days, range 7-1824, Fig.1). However, no variable predicted pair
bond duration (R2 = 0.24, F3,26 = 2.67, group size z = 0.30±0.17, t = 1.72, p = 0.10, territory
quality z = 0.07±0.05, p = 0.18, and relatedness z = 2.70±2.10, p = 0.21).
Outcomes of divorce
When incestuous pairs divorced, both pair-members subsequently paired with a less related
individual (LR estimator for females: x = -0.50, males: x = -0.54, both p = 0.01, Table 3).
Genetic relatedness between the bird that dispersed and the new partner were not different
from the average genetic relatedness between the same dispersing bird and any other
dominant bird of the opposite sex in the population (paired t-test for divorced females that
dispersed, n = 15, t = -0.09±0.01, df = 10, p = 0.47). For non-incestuous birds, there was no
evidence of an effect of divorce on the genetic relatedness with their new partner (Table 3,
females t = 0.02, df = 31.99, p = 0.59; males t = -0.03, df = 26.74, p = 0.33). No improvement
in reproductive success followed divorce in either incestuous or non-incestuous pairs (Table
3). Males that were divorced by a non-incestuous partner ended up with a partner that was
older than the previous one (x = 17 months older; df = 9.95, p = 0.02, Table 3). No other
129
partner quality attributes differed before and after divorce (Table 3). Annual reproductive
success of divorcees with their new partners was not significantly different between the sexes
(n = 45 pairs, x = 0.57±0.98 and 0.27±0.65 fledgling/year for female and male respectively, z
= 0.41, df = 32.90, p = 0.07), regardless of whether females or males stayed.
On average, females dispersing after divorce moved to territories with more Pandanus
cover (i.e. to a higher quality territory; n = 44, t = 2.20±1.06, df = 34, p = 0.01; Fig.3), but
males did not (n = 7, x = -3.00, p = 0.98). For both sexes, post-divorce dispersal did not lead
to a change in group size and the direction of the effect was opposite from expected (i.e.
individuals moved to smaller groups; females x = -1.43, df = 34, p = 0.26; males x = -2.25, p
= 0.10). For three divorcing females, we observed fights between the replacing and the
dispersing female, followed by the intruder female taking over the breeding position. The
replacing females were an older dominant female, an immigrant of unknown age and a
subordinate female and all of those three females were in neighboring territories before
usurping the dominant position. Afterwards, the evicted females were observed floating
through several territories before taking new breeding positions. All three evicted females
moved to an inferior territory, with lower Pandanus cover. More generally, when divorcing
females moved to an inferior territory after divorce (an indication that they may have been
evicted), 67% of the ex-partners (10 of 15) were joined by an older female. In contrast, when
divorcing females moved to a better territory, 30% of the divorced males (8 of 27) obtained
an older female, which is significantly less often (8/27 vs 10/15, χ2 = 5.40, df = 1, p = 0.02).
On average, the time to obtain a new breeding position after divorce was longer for females
that moved to lower-quality territories (n = 8, median= 42 days) than for females that moved
to higher-quality territories (n =12, median = 6 days, data from 2005-2010 only), although
those differences were not significant (t-test of log-transformed days to re-pair after divorce, t
= -1.76, df = 18, p=0.09). Widowed females that dispersed did not obtain territories of better
130
quality (Fig.3, paired t-test, n = 13, x = -0.38, df = 12, p = 0.24), or experience a change in
group size (x = -0.83, df == 12, p = 0.28). No other characteristic was different when
comparing divorcees and widows (Table S3).
Discussion
We found support for several hypotheses for divorce in cooperatively breeding M. coronatus
(Table 1). The Inbreeding Avoidance hypothesis was supported by the observation of high
rates of divorce among incestuous pairs and that those individuals paired with less related
partners after divorce. The Better Option hypothesis was supported by evidence that
dispersing females obtained higher quality territories following divorce. We also directly
observed a few evictions of breeding females by a usurper and some further indirect support
for the Forced Divorce hypothesis. Finally, we found that the occurrence of infidelity (EPP)
was a predictor of divorce, although its meaning is unclear in terms of long-term fitness for
either sex.
Divorce to escape inbreeding in incestuous pairs
Divorce in M. coronatus is an important mechanism to escape incest (Fig.1). Incestuous pairs
frequently resulted when birds did not disperse from their natal or breeding territories (Tables
S1 and S2), suggesting constraints on opportunities for dominant and subordinate birds
dispersing to a breeding position. In line with this, many incestuous pairs persisted for a
substantial amount of time (average = 107 days, max = 1 year). Since males are more
philopatric than females, mother-son pairings were the most common form of incestuous
pairs, and the risk of engaging in an incestuous relationship following death or dispersal of a
partner was three times greater for females (25% females vs 8% males). Females usually
dispersed after divorce, but were less likely to disperse after death of a partner (Table S2),
131
despite the high risk of incest. This lack of breeder dispersal after an unexpected event, such
as partner death, implies that divorce in M. coronatus is usually an opportunistic stay-and-
wait strategy, and breeder dispersal is triggered when a more attractive vacancy is created
elsewhere by the death or divorce of a breeder.
Incest is costly for females but even more so for males: incestuous pairs have a 30%
reduction in hatching success, and almost 37.5% of the surviving hatchlings are sired by an
EP male (compared to 3.3% EPP in non-incestuous pairs, Kingma et al. 2013). Nonetheless,
it was usually females that dispersed: 15/17 incestuously paired females dispersed. The fact
that many incestuous pairings did not dissolve immediately (see above) suggests that females
moved voluntarily, rather than being coerced or evicted by the male: if males were evicting
an incestuous partner, given their low expectations of reproductive success (low hatching
success and high EPP), we would have expected eviction to occur more rapidly.
Divorce driven by incest avoidance is not uncommon in cooperative breeders, as they
are usually constrained by a social environment with a high density of relatives (Koenig and
Haydock 1998, Daniels and Walters 2000, Hatchwell et al. 2000, Cockburn et al. 2003). In
our study, divorce is a straightforward solution to end incest, and its outcome benefits both
members: males and females paired with a less related individuals after divorce (Table 3).
However, we found no evidence that dispersers actively sought unrelated partners, since the
degree of relatedness among newly formed pairs could have been generated by random
mating.
Divorce and EPP: infidelity in non-incestuous pairs
Despite the apparent difficulties of changing breeding partners, divorce was not restricted to
incestuous pairs. However, in non-incestuous pairs, genetic relatedness did not predict
divorce. Rather, divorce was related to female infidelity: 21% of divorcing pairs had EPP in a
132
previous brood compared to 8% in non-divorcing pairs (Fig. 2). The link between divorce and
previous occurrence of EPP in non-incestuous pairs might reflect two alternative scenarios:
EP mating is used to prospect for future partners (i.e. looking for a better quality partner)
(Spoon et al. 2007, Pérez-Staples et al. 2013) or it reflects weak pair bonds (Forstmeier et al.
2014). Our available evidence provides no support for EP mating as a prospecting mechanism
for a partner of better quality: only one EP father (out of 14) became the new partner after a
divorce. In this instance, a female left a son and paired with a previous EP partner who was
paired to his daughter until then. We suggest that the positive association between divorce
and EP paternity indicates that divorce and EP mating serve – at least to some extent – the
same purpose. This relationship reflects the intense limitations on social partner choice
(Cézilly and Nager 1995, Ramsay et al. 2000), whereby EP mating is an interim solution for
females to compensate for a suboptimal partnership until this can be rectified by acquiring a
new social partner through divorce (as in Ramsay et al. 2000).
Reproductive success is not a driver or outcome of divorce
We expected divorce to lead to an increase in reproductive success, or to occur as a response
to reproductive failure (Ens et al. 1993, Dhondt and Adriaensen 1994, Black et al. 1996,
Culina et al. 2014). However, divorce in non-incestuous pairs of M. coronatus is not
obviously driven by reproductive failure: neither number of fledglings nor hatching success
predicted divorce (Table 2). Similarly, there was no change in reproductive success after
divorce (Table 3). More importantly, divorcees did not obtain higher reproductive success
compared to individuals that had changed partner due to partner death (Table S3). These
patterns could be due to the fact that fairy-wrens are fairly long-lived (Cockburn et al. 2003):
improvements in reproduction for species with long-term permanent pair bonds often occur
slowly with pair bond duration, as a consequence of mate familiarity or breeding experience
(Marzluff and Balda 1992, Cézilly et al. 2000, Van de Pol et al. 2006, Sánchez-Macouzet et
133
al. 2014). Thus, benefits of divorce might become evident only after multiple breeding
attempts: indeed, more generally, reproductive benefits of divorce in long-lived species are
less evident (Culina et al. 2014). Additionally, environmental stochasticity (Culina et al.
2014), the proposed main explanation for why performance at later breeding stages does not
predict divorce, could apply to breeding success as a whole for M. coronatus. Reproductive
success for this species is largely determined by predation and flooding events (unpubl. data),
as for many tropical species (Martin 2015). Such high environmental stochasticity reduces
the potential for reproductive failure to be selected for as a driver of divorce, as well as
hampering detection of reproductive failure driving individual divorce decisions.
Alternatively, divorce might function as a mechanism to obtain long-term benefits, rather
than an immediate reproductive benefit, as we will discuss below.
Improvement after divorce: territory quality over partner quality?
Divorced females that dispersed usually obtained a territory of higher quality. This was
independent of whether females divorced a related or unrelated partner or other potential
benefits such as number of subordinates on the territory. Such an improvement in territory
quality did not occur in dispersing widowed females (Fig. 3) indicating that obtaining a better
quality territory is a clear benefit associated with divorce rather than breeder dispersal per se.
The timing of such divorces was most likely opportunistic, with dominant females seizing an
opportunity to obtain a better breeding territory. This seems to be an important female
strategy as a third of female breeding vacancies (41/119, Table S2) were filled by another
dominant female dispersing, despite the constant presence of subordinate females in the
population. The quality of territories has long-term importance for species with site fidelity
(Desrochers and Magrath 1996, Heg et al. 2003). This is especially true for M. coronatus, in
which territories are defended year-round (Hall and Peters 2008) and breeding vacancies
134
occur rarely. Furthermore, higher quality territories are associated with long-term benefits,
because they experience greater nest success (unpubl. data) and support larger social groups,
with more subordinate helpers (Kingma et al. 2011a), and helpers over time have a positive
effect on fledgling production and breeder’ survival (Kingma et al. 2010). Although
divorcing females did not immediately acquire a larger group –presumably because
subordinates often disperse immediately after death of a related dominant (unpublished data)
- in the long run a high quality territory is important for reproductive success. Therefore, in
species with site fidelity, territory quality could take priority over partner quality, and divorce
could be considered indicative of territory choice rather than partner choice (Ens et al. 1993,
Desrochers and Magrath 1996).
Divorce: mostly a female-driven strategy?
In agreement with comparative evidence (Culina et al. 2014), our evidence shows that
females benefit more from divorce. First, most commonly the female instigated divorce by
leaving (88% of divorces, Table S2) and this was beneficial: divorcees that did not disperse
were likely to end up in an incestuous partnership with a resident, former subordinate male
(Table S1). Moreover, females dispersing after divorce acquired a higher quality territory
(Fig. 3), but no male dispersed to a better territory (n = 7, x = -3.00, p = 0.98). Additionally,
some divorces may have been triggered by female-female competition (i.e. forced divorce;
Taborsky and Taborsky 1999, Jeschke et al. 2007), driven by older females evicting younger
ones. Direct evidence for this hypothesis is limited: we observed that three divorces were
preceded by fights with a usurper female, and afterwards the evicted females dispersed to
lower quality territories. More broadly, we found that those females that dispersed to a lower-
quality territory after divorce took longer to re-pair and were more likely to be replaced by an
older female than females dispersing to higher-quality territories (Table 3). Nevertheless, we
135
cannot discard the possibility that such patterns reflect an undetected process of partner
choice by males accepting older females.
Conclusion
We found that several non-exclusive hypotheses simultaneously explained divorce in this
tropical cooperative breeder (Table 1). By adding demographic covariates and differentiating
pairs according to their degree of relatedness, we found that divorce is a strategy to escape
inbreeding for incestuous pairs. For unrelated pairs, previous female infidelity (presence of
EP paternity) predicted divorce even though this was not directly linked to future pairing with
the EP sire. Additionally, females appear to use divorce as a strategy to acquire a better
quality territory. The importance of the acquisition of a better quality territory is in agreement
with the absence of immediate improvements in reproductive success, and the species’ slow
life-history: annual survival is high, annual reproductive success is generally low with a large
stochastic component, and the benefits of a good territory are expected to accrue slowly over
the long-term. Our detailed longitudinal data and reliable information on survival and
movement of all breeders allowed us to overcome the most common limitations that hamper
the understanding of divorce (see introduction). Our study provides a clear illustration of the
complexity of factors underlying divorce for species with continuous partnerships and year-
round territoriality.
Funding
Funding was provided by a Max Planck Society Minerva Fellowship, an ARC Future
Fellowship (FT 110100505) and a Monash University Research Accelerator Program award
to A.P. and by The Holsworth Wildlife Research Endowment-Equity Trustees Charitable
Foundation, Stuart Leslie Bird Research Award-Birdlife Australia and a Monash University
Research and Graduate Scholarship to N.H.A.
136
Acknowledgements
We thank K. Delhey and N. Teunissen for constructive comments on this manuscript. Birds
were banded following protocols of the Australian Bird and Bat Banding Scheme (ABBBS
license 2230 to AP). Protocols were approved by the ethics committees of the School of
Biological Sciences at Monash University (BSCI/2011/28) and the Max Planck Institute for
Ornithology under national and international legislation for animal use in research. All
research was approved by the Western Australian Department of Environment and
Conservation and by the Australian Wildlife Conservancy (AWC).
Data Accessibility
Analyses reported in this article can be reproduced using the data provided by Hidalgo
Aranzamendi et al. (2016).
References
Andersson, M. 1994. Sexual Selection. Princeton University Press. USA.
Barrett, E. L. B., W. Boner, E. Mulder, P. Monaghan, S. Verhulst and D. S. Richardson.
2012. Absolute standards as a useful addition to the avian quantitative PCR telomere
assay. Journal of Avian Biology 43:571-576.
Bates, D., M. Maechler, B. Bolker and S.Walker. 2015. Fitting Linear Mixed-Effects Models
Using lme4. Journal Statistical Software 67: 1-48.
Barton, K. 2014. MuMIn: Multi-Model Inference. R v1.12.1, http://cran.r
project.org/web/packages/MuMIn/index.html
Birkhead, T. R. and A. P. Møller. 1992. Sperm Competition in Birds. Evolutionary Causes
and Consequences. Cambridge (UK): Academic Press Inc.
Black, J. M. 1996. Partnerships in Birds: the study of monogamy. Oxford (UK). Oxford
University Press.
137
Blondel, J., P. Perret and M. J. Galan. 2000. High divorce rates in Corsican blue tits: how to
choose a better option in a harsh environment. Oikos 89:451-460.
Cardoso, G. 2011. Paradoxical calls: the opposite signalling role of sound frequency across
bird species. Behavioral Ecology10.1093/beheco/arr200.
Cézilly F, Nager RG. 1995. Comparative evidence for a positive association between divorce
and extra-pair paternity in birds. Proc Biol Sci. 262:7-12.
Cézilly, F., M. Préault, F. Dubois, B. Faivre and B. Patris. 2000. Pair-bonding in birds and
the active role of females: a critical review of the empirical evidence. Behavioral
Processes 51:83-92.
Choudhury, S. 1995. Divorce in birds: a review of the hypotheses. Animal Behaviour 50:413-
429.
Cockburn, A., H. L. Osmond, R. A. Mulder, D. J. Green and M. C. Double. 2003. Divorce,
dispersal and incest avoidance in the cooperatively breeding superb fairy-wren (Malurus
cyaneus). Journal of Animal Ecology 72:189–202.
Cockburn, A, H. L. Osmond and M. C. Double. 2008. Swingin’ in the rain: condition
dependence and sexual selection in a capricious world. Proceedings of the Royal Society
of London B. 275: 605-612.
Cox, D. R. 1972. Regression models and life-tables. Journal of the Royal Statistical Society
B Methodologies. 34:187-220.
Culina, A., R. Radersma and B. C. Sheldon. 2014. Trading up: the fitness consequences of
divorce in monogamous birds. Biological Reviews 90:1015-1034.
Culina, A., S. Lashish and B. C. Sheldon. 2015. Evidence of a link between survival and pair
fidelity across multiple tit populations. Journal of Avian Biology 46:507–515.
Daniels, S. J. And J. R. Walters. 2000. Between-year breeding dispersal in red-cockaded
woodpeckers: multiple causes and estimated cost. Ecology 81:2473-2484.
138
Davies, N. B. 1989. Sexual conflict and the polygamy threshold. Animal Behaviour 38:226-
234.
Desrochers, A. and R. D. Magrath. 1996. Divorce in the European Blackbird: seeking greener
pastures? In: Black JM, editor. Partnerships in Birds. The Study of Monogamy. Oxford
(UK): Oxford University Press. p. 235–248.
Dhondt, A.A. and F. Adriaensen. 1994. Causes and effects of divorce in the Blue tit (Parus
caeruleus). Journal of Animal Ecology 63:979–987.
Diez, D. M. 2012. OIsurv: Supplement to survival analysis tutorial. R v1.0. http://CRAN.R-
project.org/package = OIsurv
Dubois, F., F. Cézilly and M. Pagel. 1998. Mate fidelity and coloniality in waterbirds: a
comparative analysis. Oecologia 116:433–440.
Ens, B. J., S. Choudhury and J. M. Black. 1996. Mate fidelity and divorce in monogamous
birds. In: Black JM, editor. Partnerships in Birds. The Study of Monogamy. Oxford
(UK): Oxford University Press. p. 235–248.
Ens, B. J., U. N. Safriel and M. P. Harris. 1993. Divorce in the long-lived and monogamous
oystercatcher (Haematopus ostralegus): incompatibility or choosing the better option?
Animal Behaviour 45:1199-1217.
Forstmeier, W., S. Nakagawa, S. C. Griffith and B. Kempenaers. 2014. Female extra-pair
mating: adaptation or genetic constraint? Trends in Ecology and Evolution 29:456-464.
García-Navas, V. and J. J. Sanz. 2011. Females call the shots: breeding dispersal and divorce
in blue tits. Behavioral Ecology 22:932-939.
Green, D. J., E. A. Krebs and A. Cockburn. 2004. Mate choice in the Brown Thornbill
(Acanthiza pusilla): are settlement decisions, divorce and extra-pair mating
complementary strategies? Behavioral Ecology and Sociobiology 55:278-285.
139
Hall, M. L. and A. Peters. 2008. Coordination between the sexes for territorial defence in a
duetting fairy-wren. Animal Behaviour 76:65-73.
Hall, M. L., S. A. Kingma and A. Peters. 2013. Male songbird indicates body size with low-
pitched advertising songs. PLOS One 8:e56717.
Hatchwell, B. J., A. F. Russell, D. J. Ross and M. K. Fowlie. 2000. Divorce in cooperatively
breeding Long–tailed Tits: a consequence of inbreeding avoidance? Proceedings of the
Royal Society of London B 267:813-819.
Heg, D., B. J. Ens, R. T. Burke, L. Jenkins, J. P. Kruijt. 1993. Why does the typically
monogamous oystercatcher (Haematopus ostralegus) engage in extra-pair copulations?
Behaviour 126:247-288.
Heg, D., L. W. Bruinzeel and B. J. Ens. 2003. Fitness consequences of divorce in the
oystercatcher (Haematopus ostralegus). Animal Behaviour 66:175-184.
Hidalgo Aranzamendi, N., M. L. Hall, S. A. Kingma, P. Sunnucks and A. Peters. 2016. Data
from: Incest avoidance, extra-pair paternity and territory quality drive divorce in a year-
round territorial bird. Behavioral Ecology. http://dx.doi.org/10.5061/dryad.bh526
Jeschke, J. M., S. Wanless, M. P. Harris and H. Kokko. 2007. How partnerships end in
guillemots (Uria aalge): chance events, adaptive change, or forced divorce? Behavioral
Ecology 18:460-466.
Johnson, P. C. D. 2014. Extension of Nakagawa and Schielzeth’s R2 GLMM to random
slopes models. Methods in Ecology and Evolution 10.1111/2041-210X.12225.
Johnston, V. H. and J. Ryder. 1987. Divorce in larids: a review. Colonial Waterbirds 10:16–
26.
Kalinowski, S. T., M. L. Taper and T. C. Marshall. 2007. Revising how the computer
program CERVUS accommodates genotyping error increases success in paternity
assignment. Molecular Ecology 16: 1099-1106.
140
Kempenaers, B., F. Adriaensen and A. A. Dhondt. 1998. Inbreeding and divorce in blue and
great tits. Animal Behaviour 56:737-740.
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology. 9:10.1186/1472-6785-9-15.
Kingma, S. A., M. L. Hall, E. Arriero and A. Peters. 2010. Multiple benefits of cooperative
breeding in Purple-crowned Fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79:757-768.
Kingma, S. A., M. L. Hall and A. Peters. 2011a. No evidence for offspring sex-ratio
adjustment to social or environmental conditions in cooperatively breeding purple-
crowned fairy-wrens Behavioral Ecology and Sociobiology 65:1203-1213.
Kingma, S. A., M. L. Hall and A. Peters. 2011b. Multiple benefits drive helping behaviour in
a cooperatively breeding bird: an integrated analysis. The American Naturalist 177:486-
495.
Kingma, S. A., M. L. Hall and A. Peters. 2012. Sperm storage reflects within- and extra-pair
mating opportunities in a cooperatively breeding bird. Behavioral Ecology and
Sociobiology 65:1203–1213.
Kingma, S. A., M. L. Hall and A. Peters. 2013. Breeding synchronization facilitates extra-
pair mating for inbreeding avoidance. Behavioral Ecology 10.1093/beheco/art078.
Koenig, W. D. and Haydock J. 1998. Incest and Incest Avoidance. In Koenig WD and
Dickinson J, editors. Ecology and Evolution of Cooperative Breeding in Birds.
Cambridge (UK) Cambridge University Press, p 142-156.
Kuznetsova, A., P. B. Brockhoff and R. H. Bojesen Christensen. 2014. LmerTest: Tests for
random and fixed effects for linear mixed effect models R v2.0-29. http://CRAN.R-
project.org/package = lmerTest
141
Ludwig, S. C. and P.H. Becker. 2006. Waiting for the mate? Spatial behaviour of common
terns (Sterna hirundo) during courtship. Animal Behaviour 72:1093-1102.
Lynch, M. and K. Ritland. 1999. Estimation of pairwise relatedness with molecular markers.
Genetics 152:1753–1766.
Marshall, T. C., J. Slate, L. E. B. Kruuk and J. M. Pemberton. 1998. Statistical confidence for
likelihood-based paternity inference in natural populations. Molecular Ecology 7:639-
655.
Martin, T. E. 2015. Age-related mortality explains life history strategies of tropical and
temperate songbirds. Science 349:966-969.
Marzluff, J. M., and R. P. Balda. 1992. The Pinyon Jay: Behavioral ecology of a Colonial and
Cooperative Corvid. London (UK): A and C Black Publishers.
McNamara, J. M. and P. Forslund .1996. Divorce rates in birds: Predictions from an
optimization model. The American Naturalist 147:609-640.
Nakagawa, S. and H. Schielzeth. 2010. Repeatability for Gaussian and non-Gaussian data: a
practical guide for biologists. Biological Reviews Cambridge Philosophical Society
85:935-56.
Nakagawa, S. and H. Schielzeth. 2013. A general and simple method for obtaining R2 from
generalized linear mixed-effects models. Methods in Ecology and Evolution 4:133-142.
Orell, M., S. Rytkonen and K. Koivula. 1994. Causes of divorce in the monogamous Willow
tit (Parus montanus) and consequences for reproductive success. Animal Behaviour
48:1143-1154.
Otter, K. and L. Ratcliffe. 1996. Female initiated divorce in a monogamous songbird:
abandoning mates for males of higher quality. Proceedings of the Royal Society of
London B 263:351-355.
142
Peakall, R., P. E. Smouse. 2012. GenAlEx 6.5: Population genetic software for teaching and
research-an update. Bioinformatics 28:2537-2539.
Pérez-Staples, D., M. Osorio-Beristain, C. Rodríguez and H. Drummond. 2013. Behavioural
roles in booby mate switching. Behaviour 150:337-357.
Peters, A., S. A. Kingma and K. Delhey. 2013. Seasonal male plumage as a multi-component
sexual signal: insights and opportunities. Emu 113:232-247.
Quinn, G and M. Keough. 2002. Experimental Design and Data Analysis for Biologists.
Cambridge (UK): Cambridge University Press.
R Core Team. 2015. R: A language and environment for statistical computing. R Foundation
for Statistical Computing, Vienna, Austria. http://www.R-project.org/.
Ramsay, S. M., K. A. Otter, D. J. Mennill, L. M. Ratcliffe and P. T. Boag. 2000. Divorce and
extra-pair mating in female black-capped chickadees (Parus atricapillus): separate
strategies with a common target. Behavioral Ecology and Sociobiology 49:18-23.
Rowley, I. and E. Russell. 1997. Fairy-wrens and Grasswrens Maluridae, Oxford (New
York): Oxford University Press.
Sánchez-Macouzet, O., C. Rodriguez and H. Drummond. 2014. Better stay together: Pair
bond duration increases individual fitness independent of age-related variation.
Proceedings of the Royal Society of London B 281: doi: 101098/rspb20132843.
Skroblin, A. and S. Legge. 2010. The distribution and status of the western subspecies of the
purple-crowned fairy-wren (Malurus coronatus coronatus). Emu 110:339-347.
Skroblin, A., A. Cockburn and S. Legge. 2014. The population genetics of the western
purple-crowned fairy-wren (Malurus coronatus coronatus), a declining riparian
passerine. Australian Journal of Zoology 62:251-259.
143
Spoon, T. R., J. R. Millam and D. H. Owings. 2007. Behavioural compatibility, extra-pair
copulation and mate switching in a socially monogamous parrot. Animal Behaviour
73:815-824.
Taborsky, B. and M. Taborsky. 1999. The mating system and stability of pairs in kiwi
(Apteryx spp). Journal of Avian Biology 30:143-151.
Van de Pol, M., D. Heg, L. W. Bruinzeel, B. Kuijper and S. Verhulst. 2006. Experimental
evidence for a causal effect of pair-bond duration on reproductive performance in
oystercatchers (Haematopus ostralegus). Behavioral Ecology 17:982-991.
Varian-Ramos, C. W. and M. S. Webster. 2012. Extra-pair copulations reduce inbreeding for
female red-backed fairy-wrens (Malurus melanocephalus). Animal Behaviour 83:857-
864.
144
Table 1: Overview of main hypotheses proposed to explain divorce in year-round territorial species with continuous partnerships, and their
support in Malurus coronatus
Hypothesis Assumption Predicted driversof divorce
Support in M. coronatus Predicted outcomes of divorce Support in M. coronatus
Better Option a,b Divorce occurswhen one memberis able to obtain abetter partner orterritory
Low breedingsuccess
No (for non-incestuous pairs) Improved reproductive success for one ofthe former members of a pair
No
High within-pairrelatedness
No (for non-incestuous pairs) Improvement of genetic relatedness forboth ex-partners after repairing
No
Inferior territory orpartner
No – territory qualityNo – group size
The dispersing pair member obtains:-better territory- larger groupOne of the ex-partners finds a partner ofhigher quality
YesNoNo f
InbreedingAvoidance c
Partners divorceand one dispersesto avoid breedingwith philopatricoffspring
Incest Yes (for incestuous pairs) Improvement of genetic relatedness forboth ex-partners after repairingImprovement of reproductive success forboth ex-partners
Yes
No
Forced Divorce d Divorce causedby intrusion of athird individual
Traits of thirdindividual (newpartner)
Some evidence for eviction byolder females g
Intruder benefits No
Evicted bird suffers Some evidence, somefemales obtain worseterritories
Alternativemating strategy e
Infidelityassociated withdivorce
Extra-pair paternity(EPP)
Yes (for non-incestuous pairs) Female pairs with EP fatherMales evicts unfaithful female
NoNo
aSee also Choudhury (1995) for detailed descriptions of hypotheses and other hypotheses with overlapping assumptions or not applicable to year-roundterritorial species with continuous partnerships. bDavies 1989, Ens 1993, and Magrath 1996 c Koenig and Haydock 1998, Hatchwell et al. 2000, Cockburn etal. 2003 dTaborsky and Taborsky 1999, eBirkhead and Møller 1992. f Males re-pairing with an older female is presumed indication of eviction of the ex-partner by new female, since males do not disperse to mate with an older female. g Definitive evidence requires direct behavioral observations of the processof divorce
145
Table 2: Drivers of divorce in non-incestuous pairs: effects of incompatibility (relatedness), reproductive success (number of fledglings,
hatching success), previous extra-pair offspring, group size and territory quality, on the probability of divorce (binomial GLMMs; controlling for
female origin). Model (a) test whether number of fledglings predicted the probability of divorce, including pairs that were never successful (n =
126 pairs, 220 observations) and; Model (b) test whether number of fledglings predicted the probability of divorce and included only pairs that
produced at least one genotyped offspring so that information on extra-pair paternity (EPP) was available (n = 53 pairs, 81 observations); model
(c) tests whether reduced hatching success predicted the probability of divorce, and included only pairs for which hatching success data was
available (n = 24 pairs, 44 observations). Random effects did not contribute to the variance explained by the models.
(a) (b) (c)Explanatory variables Estimate se z value Estimate se z value Estimate se z valueWithin-pair relatedness (LR estimator) 2.03 2.51 0.81 4.97 5.03 0.99 3.54 4.84 0.73Number of fledglings 0.01 0.19 0.03 0.11 0.23 0.48 n.a. - -Hatching success n.a - - n.a. - - -1.99 1.47 -1.35Extra-pair offspring (y/n) n.a. - - 2.56 1.11 2.30* n.a. - -Group size -0.03 0.18 -0.18 0.20 0.21 0.96 -0.10 0.18 -0.55Territory quality -0.01 0.06 -0.19 0.05 0.11 0.44 0.08 0.08 0.96Female origin (immigrant y/n) -1.58 0.90 -1.76 -1.09 1.21 -0.90 n.a. - -
*p < 0.05n.a. = term not fitted in this model
146
Table 3: Paired comparisons between the new partner and the partner before divorce (within-bird comparison) to analyze whether divorceresulted in an improvement for divorced females and males. Shown are mean of the change from former to subsequent partners of divorcedindividuals and p-values for (a) incestuous and (b) non-incestuous pairs. Positive values indicate improvement after divorce; the opposite is truefor genetic relatedness. GLMMs for variables with more than one entry per individual, random effects did not contribute to the varianceexplained by the model, n = number of individuals.
(a) incestuousFemales Malesn Mean p n Mean p
Genetic relatednessa 13 -0.50 0.01 13 -0.54 0.01Reproductive successa,c 7 0.28 0.20 7 0.31 0.17Partner age (months)a,d 6 11.43 0.31 6 6.28 0.99Partner size (tarsus length, mm)b 13 -0.43 0.29 10 0.08 0.81
(b) non-incestuousFemales Malesn Mean SE p n Mean SE p
Genetic relatedness 26 0.02 0.04 0.59 28 0.01 0.03 0.33Reproductive successa,c 12 0.09 - 0.31 14 1.00 - 0.34Partner age (months)d 15 3.31 5.14 0.53 11 13.25 4.90 0.02Partner size (tarsus length, mm) 27 -0.18 0.14 0.35 26 -0.15 0.21 0.41
a Wilcoxon test or bt-tests for variables with reduced sample size
c Reproductive success is number of fledglings of the first year with the new partner minus number of fledglings produced during the last yearwith the previous partner
d only birds with known hatch date
147
Figure 1. Pairs of first-degree relatives (incestuous) (n = 28) divorced more often and had
shorter pair bonds than non-incestuous pairs (n = 285) (χ 2 = 18.50, df =1, p < 0.001). Shown
is the probability that the pair bond is ongoing; vertical lines indicate pair bonds ending due
to deaths; each step down indicates a divorce.
148
Figure 2. Pairs that divorced were more likely to have had extra-pair paternity (EPP) in
previous broods compared to pairs in which one or both partners died, or ongoing
relationships (z = 2.56±1.11, p = 0.02). Bars show percentage of pairs with at least one extra-
pair offspring in previous broods; numbers indicate sample size (total number of pairs, non-
incestuous pairs only).
149
Figure 3. On average, females that dispersed after divorce (panel (a); n = 44) moved to better
quality territories (T2) than their initial territory (T1; x = 2.20, p = 0.01). This contrasts with
widowed females that dispersed (panel (b); n = 13): these did not obtain a better territory (x =
-0.38, p = 0.24). Territory quality is expressed is an index of Pandanus aquaticus cover (1-
20). Grey lines connect T1 and T2 for individual females, black lines connect medians. Box
plots showing the interquartile range (box), medians and SE.
150
Supplementary Material
Table S1: Most females (90% of 142) acquired their first partner by dispersal within or immigrating from outside the study area, while most
males (80% of 132) acquired their first partner primarily on or within 2 territories of their natal territory.
Acquisition of first partner Females Males
Incestuous Non-incestuous Total Incestuous Non-incestuous Total
Inheritance 10 2 12 20 12 32
Fission - Subordinate budded partof natal territory
0 2 2 0 14 14
Local dispersal (1-2 territoriesdistance)
0 11 11 0 60 60
Dispersal > 2 territories (0.1-6 Km) 0 74 74 0 17 17
Immigrant from outside study area 0 43 43 0 9 9
151
Table S2: Females are more likely to disperse after divorce than are males. Individuals that dispersed after divorce or partner-loss usually
acquired unrelated partners. Individuals that stayed in the territory were more likely to form an incestuous pair bond.
Acquisition ofsubsequent partners
Dominance taken byNew partner for males New partner for femalesIncestuous Non-incestuous Incestuous Non-incestuous
After divorce Stayed(n = 50) Subordinate within group 6 5 3 0
Neighbour subordinate 0 15 0 3
Neighbour dominant 0 16 0 0
Immigrant 0 2 0 0Dispersed(n = 50)* 0 6 1# 43
Total individuals 6 44 4 46After death of partner Stayed
(n = 136)Subordinate within group 2 3 11 6
Neighbour subordinate 1 27 3 25
Neighbour dominant 0 25 0 14
Immigrant 0 14 0 0
Merged/fission/demoted 0 3 0 2Dispersed(n = 28) 1 11 0 16
Total individuals 4 83 14 63* In 5 pairs male and female dispersed, 1 pair divided the territory and in 2 cases a female and a male were demoted, not included. # One female
divorcing her father dispersed 15 km and paired with her brother that dispersed before she hatched.
152
Table S3: Comparison of partners after partner change depending on whether the partnership ends due to death (widows) or divorce in non-incestuous pairs. GLMM by maximum likelihood used to compare divorced relative to widowed birds; positive means/estimates indicate highervalues for divorcees. Random effects did not contribute to the variance explained by the model.
(b) non-incestuous Females Males
Widows Divorcees Mean Estimate SE p Widows Divorcees Mean Estimate SE p
Genetic relatedness 45 22 -0.01 -0.03 0.05 0.87 56 24 -0.01 -0.01 0.04 0.87
Reproductive successa 45 18 0.22 0.22 0.40 0.58 57 18 0.04 0.04 0.43 0.92
Partner age 42 22 6.88 6.49 3.56 0.07 36 17 -0.04 -0.04 3.20 0.98
Partner sizeb 45 22 0.01 0.01 0.14 0.92 53 23 0.02 0.02 0.15 0.86a Reproductive success included offspring produced during the first year with the new partner
b Partner size estimated from tarsus length
153
General discussion
Tropical birds generally have a slow development and start reproducing relatively late in
life (Martin 1996, Wikelski et al. 2003, Møller, et al. 2007). Data on ecological drivers of
life-history characteristics typical of a slow pace-of-life, have been gathered from a few
tropical species that have been studied in detail (Martin 1996, Stutchbury and Morton
2001). This is despite the fact that the tropics have the highest bird richness and
biodiversity (Stutchbury and Morton 2001). Developing a wide-ranging theory to
understand the slow pace-of-life strategy is particularly urgent in the face of the rapid
changes occurring worldwide, since tropical species appear to be highly sensitive to human
impacts (Sekercioglu et al. 2012, Chambers et al. 2013). In order to generate this wide-
ranging theory, including more information on tropical birds is necessary.
The basic characteristics of life-history (e.g. age at first reproduction, reproductive lifespan
and ageing, number and size of offspring) determine a species position within the fast-slow
life-history continuum (Ghalambor and Martin 2001, Bielby et al. 2007). One of the most
commonly used approaches is to use single traits as indicator variables (Bielby et al. 2007),
such as reproduction. Several measures of reproductive performance over the lifetime of
individuals can show differential allocation of resources by individuals and the effects that
social and environmental parameters exert on shaping reproductive success and thereby
life-history (van de Pol 2006).
In my PhD research, I focussed on a central trait of life-history: reproduction. My model
species, the purple crowned fairy-wren, is a cooperative breeder that has long periods of
parental care, high annual survival, limited natal dispersal and pervasive constraints on
mate choice (i.e. highly saturated environment, limited turnover rates of breeding vacancies
154
and low extra-pair paternity) (Hall and Peters 2009, Kingma et al. 2009, 2010, 2013,
Hidalgo Aranzamendi et al. 2016). In my thesis I tried to understand how individual
decisions affected reproduction and how their social environment not only contributed
information on selective pressures operating within a slow pace-of-life strategy, but also to
the preservation and future management of this endangered bird.
Several ecological factors were relevant for reproduction of purple-crowned fairy-wrens.
Food abundance driven by seasonality of rainfall is fundamental to determine the breeding
cycle of this species but also the survival of offspring. High rates of nest predation are a
major driver of nest failure and the large effects of highly variable climate drivers stress the
stochastic nature of reproduction in this tropical species. Family groups are essential social
factors that brought key benefits for survival of offspring and successful reproduction.
Finally, territories located in high quality habitats consistently affect reproductive success
and individual strategies. Next, I discuss the main findings of my thesis in detail.
Using recently developed methods that can deal with complex data of multi-brooded
species, I demonstrate that female purple-crowned fairy-wrens have a fine-tuned
mechanism to time breeding cycles after rainfall and that rainfall simultaneously
determines food abundance, such that breeding and food peaks are quite well matched (Fig.
1, Chapter 1). In addition to rainfall determining timing of reproduction, quantity of rainfall
prior to egg-laying has a quantitative influence on reproductive output increasing clutch
size and number of fledglings, although it has no effect in the nestling condition (Fig. 3,
Chapter 1).
The data obtained for purple-crowned fairy-wrens reinforces the idea that more, and more
detailed, records of food availability from different tropical environments are needed, in
155
order to better understand patterns of breeding for most tropical species (Grant et al. 2000,
Sekercioglu et al. 2012). This information is urgent because it would provide information if
climate-change induced mismatches between food availability and offspring production are
occurring at tropical latitudes, similar to those described for northern hemisphere
insectivores (Visser and Holleman 2001, Visser et al. 2006a, 2006b, Schaper et al. 2012).
Recent evidence has highlighted that such phenological mismatches might be occurring in
some southern hemisphere ecosystems (Brown and Shine 2006, Chambers et al. 2013).
Moreover, rainfall was identified as the only driver of phenology for this species, and this
cue might be relevant in a similar way for many other tropical species (Skutch 1985).
Unfortunately, this leaves an important gap of information, since most current identified
mismatches between birds’ reproduction and food availability are focussed on temperature
changes and global warming. Furthermore, predictions for changes in rainfall have not been
accurately modelled at a local scale (Chambers et al. 2013), and it is likely that this needs to
be solved first, in order to predict shifts or mismatches of food availability and rainfall with
high accuracy.
Rainfall is identified as an environmental pressure driving certain characteristics of life-
history of purple-crowned fairy-wrens, as their timing of nesting attempts and quantitative
output. My analysis highlights that means and extremes of rainfall and temperature have
intricate, partly opposing effects that affect reproductive success through multiple
processes: nest inundation, predation, egg hatchability and offspring survival (Fig. 1,
Chapter 2). Climate variables have sometimes antagonistic effects on the vulnerability of
nests and fledglings; for instance, variation not only affect success directly (e.g. reducing
viability of eggs and fledgling survival) but indirectly, presumably through effects on
156
trophic interactions (i.e. predator activity) (Chapter 2). A central result here is the
importance of habitat quality, in which more vegetation cover is linked to increased
reproductive success (Fig. 1, Chapter 2). These results highlight the level of information
required before effects of climate change on populations can be accurately predicted. More
importantly, by analyzing effects at each reproductive step, losses are quantified at a finer
scale (Fig. 1, Chapter 2) and I was able to identify which environmental selective force is
relevant at which reproductive stage. Besides the impact that these complexity might have
in the future of this species, it is evident that the likelihood of nest failure at each nest
attempt is high. Therefore, it is likely that pairs would try to optimize their reproductive
output over long temporal scales, throughout their lifetime. Here, two mechanisms might be
key: having multiple broods or investing in long-term partnerships (Black 2001, Oppel et
al. 2013).
One way of improving reproductive success is by fine-tuning effective behaviours against
nest losses by predation. This could be achieved maintaining long-term relationships with
the same partner, if such partnerships are strategic to improve nest defence and brood
caring (Black 1996). Despite purple-crowned fairy-wrens having the characteristics that
would fit those predictions - year-round long-term pair bonds, almost exclusive social and
genetic partnerships and high rates of nest failure - I identify limited benefits from retaining
the same partner (Chapter 3). Although the probability of nest failure decreases over the
first years with the same partner, no further benefits are detected nor is this translated into
recruitment of more fledglings (Tables 1 and 2, Chapter 3). Instead, pairs that stay together
less long have lower hatching success. This is possibly due to decreased survival of low-
quality females (Chapter 3). As a result of these findings I hypothesize that long-term
157
relationships could be a by-product of site fidelity and lack of vacancies. In fact, my results
show that both forces operate since short to medium-length relationships were associated
with high quality territories. Apparently, in lower quality territories, pair bonds end quite
soon, or low-quality pairs remain together very long due to lack of individual
competitiveness to obtain territories of higher quality. For instance, the high relative
occurrence of incestuous pairs (9% of partnerships, 3.5% of total pair bond time), despite
high costs of incest, are clear indicators that purple-crowned fairy-wrens face strong
constraints in availability of breeding vacancies that characterise cooperatively breeding
species in general (Cockburn 1998, Hatchwell and Komdeur 2000, Koenig and Dickinson
2004). Site fidelity of pair members could be then reinforced in this species by the high
value of good territories. Territories with higher vegetation cover are key in reducing the
likelihood of nest predation (Chapter 2) and recruit more fledglings, highlighting the
importance of good assets for breeders (Chapter 3). It seems then that it is beneficial to
retain dominance in a good quality territory and long-term relationships could be a side
effect of it.
Since my study found no substantive benefits of mate retention, I hypothesise that birds
should be willing to abandon a breeding position and move to a better situation if they have
the chance, and in Chapter 4, I explore why then breeders divorce. Divorce is a common
strategy in pairs socially bonded to first-order relatives (incestuous). Reproductive success
for these pairs is severely reduced (Kingma et al. 2013) and to find divorce as an escape of
incest is an expected outcome. However, divorce is not only an incest-avoidance
mechanism but also a strategy in pairs of unrelated individuals. Here, females appear to be
the main beneficiaries of divorce and they most likely initiate the dissolution of the pair
158
bond, by taking breeding positions on territories of higher quality (Fig. 3, Chapter 4).
However, no improvement in immediate reproduction is found and this challenges what is
known about the adaptive value of divorce. My findings offer a new understanding on the
fitness benefits obtained and the complex interplay of factors in the decision to leave a
partner.
In summary, studying aspects of productivity in purple-crowned fairy-wrens has shown that
this species faces important constraints in partner choice that will eventually affect
reproduction. My thesis has shown that successful reproduction is a hard-to-reach outcome
due to external environmental pressures and that long-term pair bonds are not a key to
improve productivity. Instead, large groups and territories of high quality have consistently
show up as key benefits for productivity that also drive competition for breeding vacancies.
All these characteristics could be common place in tropical species, especially for those that
occupy territories year-round and live in saturated environments. Furthermore, in this thesis
I have dealt with ecological patterns that are not common in other systems, such as year-
round reproduction, multi-brooding, high rates of nest predation and long periods of
fledgling survival. Such patterns may prevent using common existing methods in animal
ecology, since they were developed to study species from different environments that have
fixed seasonal reproductive cycles facing different selective pressures. Investigating more
tropical species will help to develop further integrated methodologies that will allow to
explore similar measures in other species. To date, we are only starting to understand which
processes affect reproductive success and which ecological and/or social pressures are
relevant for tropical birds.
159
My research also highlighted some potential threats of climate change. In here, an increase
on climate extremes (both rainfall and temperature) might have implications in the slow
reproductive benefits that purple-crowned fairy-wrens obtain at each breeding attempt (Fig.
1, Chapter 2). Hopefully my results highlighting the importance of good habitat quality for
nest success, breeder dispersal, and support for large social groups, will not only inform,
but provide support for the ongoing successful management for the habitat of this species
(Skroblin et al. 2014). Although, we might better understand the threats of ongoing climatic
change, due to a better understanding of the nature of these threats and as the duration of
the study increases; however, their occurrence cannot be avoided nor their direct effects
mitigated. Thus, due to their slow pace of reproduction of this species, and the threats of
ongoing climatic change, preservation of significant stretches of healthy habitat supporting
growing populations, is urgently needed.
In conclusion, my results exploring which ecological and social factors determine variation
of individuals’ productivity will contribute to a core aspect of life-history theory. Moreover,
this information will be particularly relevant to understand the slow pace-of-life of tropical
species; information that is key to manage threatened species.
References
Black, J. M. 2001. Fitness consequences of long-term pair bonds in barnacle geese:
Monogamy in the extreme. Behavioral Ecology 12: 640-645.
Bielby, J., G. M. Mace, O. R. P. Bininda-Emonds, M. Cardillo, J. L. Gittleman, K. E.
Jones, C. D. L. Orme and A. Purvis. The Fast-Slow Continuum in Mammalian Life
History: An Empirical Reevaluation. The American Naturalist 169:748-757.
160
Black, J. M. 1996. Partnerships in Birds: the Study of Monogamy. Oxford: Oxford
University Press.
Brown, G. P. and R. Shine. 2006. Why do most tropical animals reproduce seasonally?
Testing hypotheses on an Australian snake. Ecology 87:133-143.
Chambers, L. E., R. Altwegg, C. Barbraud, P. Barnard, L. J. Beaumont, R. J. M. Crawford,
J. M. Durant, L. Hughes, M. R. Keatley, M. Low, P. C. Morellato, E. S. Poloczanska,
V. Ruoppolo, R. E. T. Vanstreels, E. J. Woehler and A. C. Wolfaardt. 2013.
Phenological Changes in the Southern Hemisphere PLoS ONE 8: e75514.
Cockburn, A. 1998. Evolution of Helping Behavior in Cooperatively Breeding Birds.
Annual Review of Ecology and Systematics 29: 141-177.
Culina, A., R. Radersma and B. C. Sheldon. 2014. Trading up: the fitness consequences of
divorce in monogamous birds. Biological Reviews 90:1015-1034.
Ghalambor, C. and T. E. Martin. 2001. Fecundity-survival trade-offs and parental risk-
taking in birds. Science 292: 494-499
Grant, P. R., B. R. Grant, L. F. Keller and K. Petren. 2000. Effects of El Niño events on
Darwin's Finch productivity. Ecology 81: 2442-2457.
Hall M. L. and A. Peters. 2009. Do male paternity guards ensure female fidelity in a
duetting fairy-wren? Behavioral Ecology 20: 222-228.
Hatchwell, B. J. and J. Komdeur. 2000. Ecological constraints, life history traits and the
evolution of cooperative breeding. Animal Behaviour 59: 1079-1086.
Hidalgo Aranzamendi, N., M. L. Hall, S. A. Kingma, P. Sunnucks and A. Peters. 2016.
Incest avoidance, extrapair paternity, and territory quality drive divorce in a year-round
territorial bird. Behavioral Ecology arw101.
161
Kingma, S. A., M. L. Hall, G. Segelbacher and A. Peters. 2009. Radical loss of an extreme
extra-pair mating system. BMC Ecology 9: 10.1186/1472-6785-9-15
Kingma S. A., M. L. Hall, E. Arriero and A. Peters. 2010. Multiple benefits of cooperative
breeding in Purple-crowned Fairy-wrens: a consequence of fidelity? Journal of Animal
Ecology 79:757-768.
Kingma, S. A., M. L. Hall and A. Peters. 2013. Breeding synchronization facilitates
extrapair mating for inbreeding avoidance. Behavioral Ecology 10.1093/beheco/art078
Koenig, W. and J. Dickinson. 2004. Evolution and Ecology of Cooperative Breeding in
Birds. Cambridge University Press.
Martin, T. E. 1996. Life history evolution in tropical and south temperate birds: what do we
really know? Journal of Avian Biology 27: 263-273.
Møller, A. P. 2007. Senescence in relation to latitude ad migration in birds. Journal of
Evolutionary Biology 20, 750–757.
Oppel, S., G. M. Hilton, R. Allcorn, C. Fenton, A. J. Matthews and D. W. Gibbons. 2013.
The effects of rainfall on different components of seasonal fecundity in a tropical forest
passerine. Ibis 155: 464-475.
Ricklefs, R. E. and M. Wikelski. 2002. The physiology/life-history nexus. Trends in
Ecology and Evolution 17:462-469.
Şekercioğlu, Ç. H., R. B. Primack and J. Wormworth. 2012. The effects of climate change
on tropical birds. Biological Conservation 148: 1-18.
Stutchbury, B. J. M. and E. S. Morton. 2001. Behavioral Ecology of Tropical Birds.
Academic Press, San Diego.
162
Skroblin A., S. Legge, T. Webb and L. P. Hunt. 2014. EcoFire: regional-scale prescribed
burning increases the annual carrying capacity of livestock on pastoral properties by
reducing pasture loss from wildfire. The Rangeland Journal 36: 133-142.
Skutch, A. F. 1985. Clutch size, nesting success and predation on nests of Neotropical
birds. Ornithological Monographs 36:575-594.
van de Pol, M. 2006. State-dependent life-history strategies. A long-term study on
oystercatchers. PhD Thesis. University of Groningen.
Visser, M. E., F. Adriansen, J. H. van Balen, J. Blondel, A. A. Dhondt, S. van Dongen, C.
du Feu, E. V. Ivankina, A. B. Kerimov, J. de Laet, E. Matthysen, R. McCleery, M.
Orell and D. L. Thomson. 2006a. Variable responses to large-scale climate change in
European Parus populations. Proceedings of the Royal Society of London Series B
270:367-372.
Visser, M. E., L.J.M. Holleman and P. Gienapp. 2006b. Shifts in caterpillar biomass
phenology due to climate change and its impact on the breeding biology of an
insectivorous bird. Oecologia 147:167-172.
Visser, M.E., C. Both and M. Lambrechts. 2004. Global climate change leads to mistimed
avian reproduction. Advances in Ecological Research 35:89-110.
Wikelski, M., L. Spinney, W. Schelsky, A. Scheuerlein and E. Gwinner. 2003. Slow pace
of life in tropical sedentary birds: A common-garden experiment on four stonechat
populations from different latitudes. Proceedings of the Royal Society of London B:
Biological Sciences 270, 2383–2388.